PTNA: news for Public Transport Network Analysis

General talk
, ,
463

@ToniE Would it be possible to include the Transport Victoria GTFS data https://opendata.transport.vic.gov.au/dataset/gtfs-schedule?

This is covered by the CC BY waiver we have at File:DTP Waiver signed.pdf - OpenStreetMap Wiki

1 Like
464

Thanks for pointing, I’ll work on that. I also received the same idea by @LittleJimmyR (not present here?)

I will publish this as AU-VIC-DTP

465

Any ideas or wishes for PTNA analysis of OSM data in that area, similar to AU-SA-Adelaide-Metro and AU-NSW-Sydney

466

I need some more time for this.

The link does not point to a gtfs.zip who’s structure is as requested/defined by the GTFS reference.
Instead this gtfs.zip includes 8 other (correct) GTFS zip files as described in their GTFS Release Notes at page 3

So, what you’ll get from PTNA are 8 GTFS feeds:

  • AU-VIC-DTP-1 for Regional Train
  • AU-VIC-DTP-2 Metropolitan Train
  • AU-VIC-DTP-3 for Metropolitan Train
  • AU-VIC-DTP-4 for Metropolitan Bus
  • AU-VIC-DTP-5 for Regional Coach
  • AU-VIC-DTP-6 for Regional Bus
  • AU-VIC-DTP-10 for Interstate
  • AU-VIC-DTP-11 for SkyBus

I need some more time to set up these 8 configs taking care to download the global gtfs.zip only once and not 8 times.

Fortunately, checking for update using “curl -sI ...” and checking for ‘last-modified’ header is OK.

P.S.: I did not check whether the 8 zips (their individual *.txt) can easily and correctly be merged and repacked …

1 Like
467

Yeah I guess similar would be good. I went through the errors reported by AU-NSW-Sydney and fixed a bunch, so that was helpful.

I don’t quite understand though in Sydney why we have the busses listed once at the top with …to be completed, but then we have the complete list further down under “Other Public Transport Lines > Bus (bus)” I tried to read the documentation but couldn’t understand how it works.

image
image489×585 29.7 KB

468

Ideally, section 2. lists all routes which exit in reality, section 3. lists all others, e.g. artefacts, routes from other areas, networks, …

The routes which exit in reality can be defined in he CSV data in the OSM wiki.
For those routes, PTNA can link route-masters and routes and perform some more checks. In addition to that, PTNA can complain if a route listed there (e.g. bus 234) is missing in OSM. Having GTFS data allows creation of the CSV list using the GTFS., there’s an ‘export’ button.

Btw. AU-VIC-DTP-* GTFS will be prepared starting 13:15 CEST UTC+02

469

Bad news: although I had success yesterday, today I get “HTTP/2 404” not found.

I’ll have to investigate whether there is a perma-link available or whether I have to parse some HTML page to extract the daily changing URL for the GTFS data.

Having a perma-link is requested in the “GTFS best practice”.

So, this is the 2nd annoyance after gtfs.zip not being a real GTFS zip but containing 8 GTFS zips.

1 Like
470

No perma-link available, I have to scan a web page for the URL.
This is nothing new for PTNA, there are some other feeds which must be handled like this.

Another start preparing AU-VIC-DTP-* at 14:30 UTC+02, 4 minutes to go

471

… done, successfully

1 Like
472

I just released a new feature for PTNA: better support of Life-cycle prefix for (public transport) route relations

You’ll see the results after the usual nightly (your local time) analysis run, starting with East-Australia in 40 minutes, ending with Alaska in 19h40m

473

This morning I added support for the analysis of public-transport data to ptna for

based on a PM by @donmac703

474

Yesterday, a new gtfs feed for public-transport data has been added to ptna

  • DE-BW-BWGesamt
    • Gesamtgebiet Baden-Württemberg
    • published by NVBW - Nahverkehrsgesellschaft Baden-Württemberg mbH on behalf of Ministry of Traffic of Baden-Württemberg
    • covering all public transport lines in this state of Germany
      • benefit: includes: regional trains, Flixbus, IC, ICE, TGV, … in and through this area
      • draw-back: many routes with identical ‘ref’: e.g. 23 times bus number “1” in different locations
1 Like
475

I just added analysis support of public-transport data in OSM to ptna for

There’s more work in the pipeline:

476

analysis of gtfs data has been added to ptna for

477

I just added support for the analysis of public-transport data in osm to ptna for

2 Likes
478

I just added analysis support of public-transport data of OSM to ptna for

479

I just added the support of the analysis of gtfs data to ptna for

@KevinMapsThings

1 Like
480

I just release the first version of a new check in ptna for public-transport data of osm

The implementation is based on a suggestion from and discussion with @donmac703

Short:

  • the distance of the first platform and the first way in a PTv2 route relation will be measured
  • the distance of the last platform and the last way in a PTv2 route relation will be measured

If the distance exceeds 20 meters, an error message will be shown.

Long:

An error message could look like:

  • PTv2 route: first way (‘Lambton Parade’ 173120433 (iD, JOSM)) is not near the first platform stop (‘Pacific Hwy before Macquarie St’ 5203972847 (iD, JOSM)). Distance = 6259 meters.

    • seen the route and first platform in PTNA

  • PTv2 route: last way (‘Pacific Highway’ 1257758004 (iD, JOSM)) is not near the last platform stop (‘Wyong Station, Stand B’ 6763249795 (iD, JOSM)). Distance = 43 meters.

    • see the route and last platform in PTNA

There are many reasons for such an error:

  • the ways in the route relations are not well sorted
    • e.g. a way from the middle has been moved to the end of the list, accidentally
      • fix this using JOSM, before using Relatify for further edits. Relatify will use the last way in the relation as the end of the route, removing other ways.
  • the platforms are not well sorted
  • the list of ways is not complete
    • e.g. missing ways at the end where the actual path to the platform in the area of a bus station is unknown
  • the list of the platforms is not complete

@donmac703 might add more details and background information.

The code currently measures the distance only between platform nodes and nodes of the way. Support for platform ways/areas and relations will come soon.
Support for the measurement of the distance between the/each platform node and the lines/segments between two nodes of the way will come later on. The LeafLet function: pointToSegmentDistance needs to be implemented in Perl.

The new code runs with the next nightly analysis. If you want to try that before: go ahead and use the button “Request start of new analysis”.

Feed-back is highly welcome

id #relatify josm

1 Like
481

Good feature (and I will probably have a lot of inconsistencies to resolve in the transport lines I maintain) but as I expected you will get also some false positive. I mean there are some configurations where the “platform” may be a little bit further to the stop position as usual, it is not necessary an “error”. Take this example: the platform is set to one corner of the transport shelter (which is common). The way and stop position are 21 meters further. Maybe the view with the imagery speaks for itself:

image
image422×359 63.5 KB

It may be difficult to close the gap in this case without inserting all parking spaces for the bus (which do not have a fix position - first come first serve)

482

The discussion behind this led to a “default” of 20 metres being established. I am uncertain if Toni has had the time to write the configurability of it yet.
However, you are correct, there are always going to be false positives, the real question is whether there is a significant number or not. Hence releasing the feature into the wild for feedback as a minimum viable product.

In my case there are approx 1400 gtfs routes (routes, not gtfs trips!) that I needed to analyze, and the problem was that I couldn’t easily (i.e. by viewing the summary page) find the ones that needed the “most” attention.

At the other end of the scale, I have one route where the distance is 23140 metres!

There are also some more “peculiarities” that I have discovered when reviewing the new PTNA analysis summary, so other improvements are possible.

The current state should be sufficient to click the relatify link and not have it confused by “false” start and end ways (if you have addressed the distant ones).