In May 2016, TreinTramBus and iRail successfully finished a crowd-funding campaign called Spitsgids, in which we were able to raise €4140 from 121 believers. We promised to hire students during open Summer of code 2016 and to open up the data through the iRail API. Today we are proud to announce the beta launch of occupancy rates in iRail!
What does a beta launch mean?
We have launched the API in production, and apps can integrate the score and can provide us with feedback. The feedback is opened up as open data in realtime under a CC0 license at https://api.irail.be/logs. For the time being, we have 4 occupancy levels instead of 3: high, medium, low and… unknown. Unknown means that we would like to invite the user to give us feedback about this train as we could not make a prediction at this time. We will keep the ‘unknown’ until we have gathered enough data. In the meantime, everyone can start using and reusing the data!
What features are we still going to implement?
Two important features still need implementation. To start with, we want to ‘transfer’ the occupancy from user feedback to the next connections on the line, based on assumptions about how occupancy evolves in a train. We already have a theoretical framework about how this should function, but the technical stuff still needs to be done.
Aside from that, and even more important, the API should learn to make predictions from user feedback. From that, it could learn about structurally busy trains, but also about more exceptional events, like festivals or when the weather is nice and everybody wants to go to the sea. This would be the real power of crowd-sourced information, and it would enable people to plan their journeys in more comfort.
If you think you can help with that, do not hesitate! Also, the more feedback you give about your train, the better!
Arne, Stan, Serkan and Pieter
We have updated the API docs slightly over here:
- Vehicle information now contains much more data about arrivals, delays, cancelations, and so forth. It’s kept backward compatible, but the new properties should allow you to make richer application views
- https is now the default. http wil keep working though, again for backward compatibility
- More alternatives added for the stations list
If you have any feedback on the documentation please do let us know!
In our next update somewhere next week, we should also be able to show occupancy estimations of trains and POST feedback on this to our API. You can follow updates on that over here: https://github.com/iRail/iRail/pull/191. We are working on this with students of Spitsgids, thanks to https://spitsgids.be.
On top of the iRail API, many requests happen. Even if we give concrete numbers, it remains difficult to understand how many requests we actually get. Serkan Yildiz, one of the open Summer of code students, created the open iRail map, that visualizes requests in real time. Check it out:
With our API’s query logs, with a ground truth by train specialists in Belgium, and with user feedback, we can show you how busy your train is. The only thing lacking to realizing this feature that the Dutch already have for years, is your support! Go to spitsgids.be and get a seat on your next journey.
— Maxime Delrue (@mdelrue) April 18, 2016
— iRail (@iRail) April 18, 2016
Time flies! But we have great news:
— iRail (@iRail) April 13, 2016
We can now officially close the era where iRail has been an advocacy group for Open Transport Data, and instead become something we have always wanted to become: a Living Lab (and/or hackerspace) to work towards a better transport experience for travelers in Belgium. The next steps are going to entail a better collaboration with players like TreinTramBus, iMinds, the European Passenger Federation and everyone who has can help us reach our goal. A first thing is going to be announced next Monday, the 18th of April! We are quite psyched about this.
TreinTramBus and iRail have already been working together in the past, analyzing the query logs we have published in collaboration with iMinds’ Data Science Lab (my current employer). The research that has been carried out was presented at the WWW2016 conference in Canada, during the USEWOD workshop. The full paper can be downloaded here.
— Marin Dimitrov (@marin_dim) April 11, 2016