PTNA: news for Public Transport Network Analysis

NL - Nederland I just added 7 more analysis of public-transport for

  • NL-GE-* (4)
  • NL-NH-* (3)

to ptna

@spaanse @IIVQ

NL - Nederland I just added 3 more analysis of public-transport for

  • NL-NH-*

to ptna

NL - Nederland I just added 5 more analysis of public-transport for

  • NL-ZH-*

to ptna

More to come step-by-step, slowly as I’m working on a GTFS/OSM comparison solution (on trip/route level).

I just released the next version of GTFS/OSM comparison into the wild:

3 Likes

I added analysis of public-transport for

I also added gtfs analysis for

3 Likes

I added analysis of public-transport for

I also added/fixed gtfs access to GTFS feeds for

There are some additional new features for the GTFS/OSM comparison

4 Likes

I just added analysis of public-transport for

to ptna

3 Likes

I just added analysis of public-transport for

to ptna

1 Like

Today I added analysis of public-transport for

to ptna

1 Like

@spaanse In the OSM wiki there is a list of GTFS sources.
You’re referring there to the NL-OVApi.
However, for historic reasons, PTNA calls this one NL-Nationaal.

Shall I rename that in PTNA? The source (*.zip !) is the same.

To me NL-OVApi makes more sense since it names the organisation that publishes it (like most other feeds).
If you change it, could you also update the three routes that use it?

Yep, all done.

I’ll extend the list of PTNA results for NL after my vacation.

I added a new type of check for PTv2 route relations:

  • first stop_position / platform : check whether the ‘role’ includes ‘*exit_only
  • last stop_position / platform : check whether the ‘role’ includes ‘*entry_only

If there is more than 1 stop_position resp. platform.

Check depends on analysis option ‘--check-sequence

4 Likes

I just added analysis of public-transport for

to ptna

Edit: 2024-07-07 15:50 UTC link to GTFS analysis added

Today I added analysis of public-transport for

to ptna

Thanks Carlos and others also for translations into Catalan language for ptna-www @ GitHub and also @ TransiFex

I just added analysis of public-transport for

and

to ptna

Thanks Carlos also for the pull requests #54, #55, #56 and #57: translations into Catalan language for ptna-www @ GitHub

PTNA is currently undergoing major renovations:

Q: Why all this?
A: PTNA reaches / exceeds the fair use limit for Overpass API queries

Q: Will my ‘network’ / analysis be affected?
A: Almost all ‘network’ will be changed. But only “almost all”.

Q: Will I see differences?
A: No, at least that is the goal. But: Queries to Overpass and queries to ‘osmium tags-filter’ are different. So there might be some difference. Please report any strange differences. Fine-Tuning will take place then.

Q: When will it start?
A: Tonight we start with parts of Germany in the UTC+01 time zone: DE-BY-CBB (already available), DE-BY-RVO and DE-NW-VRR. I hope it will run without any problems.

Q: Where can I see the current status of the switchover?
A: The statistics page has 3 new columns under “Filter Planet Extract” and a time zone specific output (see: UTC+01)

Thanks to @Jochen_Topf for the discussions on ‘osmium’ during the FOSSGIS OSM Community Meeting 2024 number 21

4 Likes

I assume you considered and rejected the idea of running a private Overpass server, just for PTNA – that would avoid needing to change the queries, at the cost of having to set up a full overpass server, and keep it populated with data.

That was my initial idea: Overpass instance in a docker container.
That would have been a big initial effort and would have required much resources (2 TB for the planet dump, postgis DB, …) on the FOSSGIS-owned server.

Jochen convinced me to work with planet dump, planet extracts.

I finally came up with the solution of working with country-specific extracts combined with their minute-updates from download.openstreetmap.fr

I will extend the list of country-specific extracts step-by-step, starting with Germany.
Those extracts are then split into smaller pieces: states, counties and/or area covered by the ‘network’ (as type=public_transport relation or so). Finally the smaller areas are filtered for relevant data and stored as .osm/.xml.

For the analysis part of PTNA, it is irrelevant where the data (as file) comes from, as long as it is XML. The better the filtering, the smaller the XML, the faster the parsing of the XML.

For DE-BY-CBB (Burghausen, a small city in Bavaria) the analysis results did not show any differences between Overpass and Planet Extracts. Let’s see whether this holds for the large analysis of DE-BY-RVO (Upper Bavaria, as relation based on admin boundary) and DE-NW-VRR (huge part of North Rhine-Westphalia, as relation based on public transport boundary).

BTW: minute-update of germany and split into pieces started 20 minutes ago.
I’ll keep an eye on it.

1 Like

Oops! And there are tons of missing data:

  • missing member nodes and member ways outside the extract. Extracts should include all nodes and ways of all relations, even if they are outside the polygon defining the extract.
  • other missing data is related to ‘osimium’ not providing a filter like [~'route'~'(bus|tram|train|subway|light_rail|trolleybus|ferry|monorail|aerialway|share_taxi|funicular)'] as Overpass does - especially the left part ~'route' (a RegEx)

Either I can fix this or it will be a KO criterion.

1 Like