Bus schedules are smoke and mirrors

Last week I wrote a post how Datagems had used TIBCO Streambase to connect to the Auckland Transport live stream. We also had downloaded all the schedules as these change periodically. One reader correctly pointed out pushing the data into a db isn't strictly streaming analytics. However, we needed to start collecting data and storing it somewhere so we could do get started on some descriptive analytics which would set the basis for allowing us to create some real-time analytics or even one step further some predictive analytics.

Below are the preliminary results from just one weeks worth of bus journeys on the 926 Akoranga 8am service using TIBCO spotfire X.

If I was to get on this bus, it happens at either stop 23,24 or 25 (remember my passion for perpetual motion). We can see from this table it is consistently late. What we can also see is the schedule is tweaked to give the bus service the opportunity to "catch up" and meet the on time target. However, this tweaking of this intra stop is not ideal. Especially if you are using the Auckland Transport Mobile App. The app doesn't know where all the hotspots for traffic are. It doesn't know how many people are at each stop and it certainly doesn't know how the driver runs their bus.

We can see this from the app screen grab above, I know it's not the same 8am service, This is from this afternoon, what it shows is that the app does not line up with the schedule. The app is saying the bus is 1 minute away and 3 stops. According to google maps it's 1.1km which would mean the bus would have to travel non stop at 66km/hr to get to the stop on time with no pick ups and a left hand give way turn. Is this realistic?

So, coming back to my 8am service.

I have to ask myself, given that I now understand the timetable, I know the bus is late on my street 3 days out of 5, and the app is not particularly helpful in assessing bus arrival accuracy. I think I'd rather walk. I know it's going to take 19 minutes at google pace. Over the next few weeks and months, we will continue to evolve our data set and try to understand in greater detail the complexity of running a bus network. Stay tuned for more and I welcome any feedback or questions.