Statistical Analysis of Bus OnTime Data:Route 370 case study

[boxythingy]

Route 370, one of Sydney's top 10 offenders for buses running more than 20min late:


Website with the data: https://bus-shaming.com/

[andy_centralcoast]

This is actually a pretty awesome initiative. A uni student was frustrated at the lack of public reporting and transparency of bus on time running statistics, but she realised she could generate it herself from the real-time tracking and delay data that underpins apps like TripView.

She collected all the real-time bus data for 4 months, covering 3.7 million bus trips across NSW.

Over those 4 months, 32% of buses ran on time (less than 5 minutes late). Shorter bus routes, trips outside of Sydney, and Night Ride buses were the best performing.

She has plans to expand this project to train, light rail and ferry data in the future, but she joked she may have to get a lawyer if TfNSW don't like what she's doing with their real-time data feeds.

[Stu]

Additional 370 services will be operating today between Coogee & Green Square in both directions.

[Frosty]

Interesting study but I had noticed the late running services in that study many were long Loop services so that’s not surprising or there were routes that ran in peak hour only which are very likely to be late due to traffic factors. If I’m not mistaken isn’t the current offical definition for OTR for buses as 2 mins early to 6 mins late.

Short-workings between Coogee & Green Sq for Aus Day which depot is operating the trips & is there even a desto for Green Square for the 370.

[tonyp]

I've yet to watch the whole thing - her machine-gun style of delivery had me reaching for the Panadol. Good on her for trying. As soon as TfNSW finds out, they will shut down her public avenue for statistics.

It's nothing surprising. Buses have been running late in bunches of three since DGT days. Nothing will change while their passenger exchange process and lack of traffic priority remain so backward.

It's interesting that tracking app developers can access data freely. UOW obviously hasn't been able to do that for the Gong Shuttle. Is that because privates in NSW don't have tracking of their buses?

[sunnyyan]

^ It is definitely possible to get data for the Gong Shuttle as 55A&C both show up on anytrip.com.au (uses realtime data from TfNSW). Most privates do actually have tracking from what I see, with the limits being pretty much the entire intercity train network.

[sunnyyan]

^ Anyone can access the tracking data now. See https://opendata.transport.nsw.gov.au/

[tonyp]

^ It is definitely possible to get data for the Gong Shuttle as 55A&C both show up on anytrip.com.au (uses realtime data from TfNSW). Most privates do actually have tracking from what I see, with the limits being pretty much the entire intercity train network.

Thanks, that's a really good one (once it slowly opens up and reveals everything). Why aren't these aids publicised by TfNSW?

It shows the Gong Shuttle all over the shop even on a public holiday! It must be Australia Day crowds combined with single-door buses.

I will be interested to observe IWLR. That's a bit over the shop too today.

[rogf24]

Some data on 55 route.

https://bus-shaming.com/route/5457
https://bus-shaming.com/route/5456

Also, the main part of the video goes from between 15:20 to to 26:40, before that it's basically just technical details on how she collected and analysed the data (which is still interesting) and after that questions and answers.

[Tonymercury]

Why aren't these aids publicised by TfNSW?

they were, even here.

see - https://www.computerworld.com.au/articl ... avel-data/

[tonyp]

Some data on 55 route.

https://bus-shaming.com/route/5457
https://bus-shaming.com/route/5456

Thanks, that's appalling, even worse than I thought from casual observation. This is a service that's rarely affected by traffic (in fact it "solved" the traffic). This is just terrible crowd-handling as a result of reasons previously outlined. I've been dipping into 55 today on Anytrip but don't want to draw too many conclusions because it's Australia Day, but I'm surprised at the number of buses running early, on top of the ones running a little late and very late and some on time. At one point two buses were running together, one on time, the other 10 minutes late! And this is only Wollongong. It must take exceptional talent to achieve results like these.

I'm also rather concerned to observe that trams on IWLR are tolerated running early, which they are occasionally. So much for relying on the journey planner.

[Linto63]

It's the old 'why do buses always come in 3?' problem. The reasons are fairly well documented and short of slamming doors in peoples faces, skipping stops or building in lots of dwell time, there is little that can be done to avoid without significantly increasing costs.

[tonyp]

It's the old 'why do buses always come in 3?' problem. The reasons are fairly well documented and short of slamming doors in peoples faces, skipping stops or building in lots of dwell time, there is little that can be done to avoid without significantly increasing costs.

https://vimeo.com/236435737
(click Watch on Vimeo)

[Linto63]

Can't see the clip, but from the title presume it is of the many times discussed passengers entering via the back door. But this is only part of the problem.

Reality is that even Bus A and Bus B both leave the starting point bang on time 10 minutes apart, if Bus A has a bad run with 3 sets of lights, and Bus B a clean run, at say 60 seconds a pop that alone will reduce the gap to 3 minutes. Add in a need for a wheel chair assist, a clueless passenger who wants directions, or an incompetent reverse parker on top of the number of passengers who have arrived to catch Bus B now boarding Bus A, and the gap can easily be further eroded.

It's all swings and roundabouts, but the best transport planner in the world isn't going to be able to overcome this without ridiculously long dwell times. Yes priority measures would help, but on many parts of the Broadway to Anzac Parade section, the 370 traverses sections that are already one lane in each direction, well unless permanent clearways are introduced, but these seem to be sacred cows.

[tonyp]

The end result of the in-service demonstration in the video was that 37 people boarded the bus through all doors in 26 seconds whoa to go compared with 35 people taking 1 min 19 seconds (79 seconds) to board an identical bus through the front door. This difference cumulatively adds up to a lot of time over a route. It can be noted that all-door loading also dampens down the adverse effects of a delayed bus having to pick up passengers who've come to meet the next bus, because they can get on the bus quite quickly.

Front door loading is one of the biggest contributing factors to losing time and consequent bunching. Traffic entanglements due to lack of priority run second as a factor. The Gong Shuttle is an excellent example demonstrating the magnitude of the problem of front door loading because it operates in an almost clinically traffic-free environment - and with relatively few traffic lights. The delays are almost entirely down to passenger exchange. We have a sort of wilful blindness about this issue in NSW.

[Frosty]

If your looking at the 370 late running it’s clear to see from the terminus at Coogee drivers often loose track of time, late from depot or last trip or their Maccas run took longer than expected. Plus the usual last minute straggler. Than going through Randwick 3-5 mins are easily lost due to the LR works, traffic lights & heavy traffic. Then along Anzac Parade with the LR works. No time can be gained along Lenthall St narrow street with roundabouts. Then during peak hour the run from Epsom Rd through Green Sq/Botany Rd is shocking. More delays at Sydney Park Rd due to Westconnex & local drainage works. King St speaks for itself traffic sewer one lane with cars moving in and out parking spots. Same for Glebe Pt Rd. With the final stretch of narrow inner west one lane streets & tight turns.

If your looking for long dwell times I find it’s UNSW, USYD City Rd & Broadway, Coogee terminus & Green Sq.

[Tonymercury]

It's the old 'why do buses always come in 3?' problem. The reasons are fairly well documented and short of slamming doors in peoples faces, skipping stops or building in lots of dwell time, there is little that can be done to avoid without significantly increasing costs.

And you left out the willingness of some operators to catch sight of the bus in front and then stay back to avoid passing!

[tonyp]

The original video seems to have disappeared off this thread, so I haven't finished watching it, but I was impressed to read about Katie Bell's background and her work with encouraging women into the computer sciences. It seems that we have been lucky enough to have a typically poor-quality State Transit bus journey attract the attention of somebody with completely different interests to apply her analytical skills to bringing information about bus service quality to public attention in an easily digestible manner. Good on her.

[moa999]

And frequency is the key part.
It doesn't matter if every 55 or 370 is running 20min late - to the pax it doesn't matter.

It's when one in 20min late, the next 10min late and the final one on time so clumping in 3s, that the passenger is upset

[Frosty]

That is very true frequency does matter but during the peak hours the 370 is timetabled to have buses on average about every 10 mins but regular users know that is not true as there are often large gaps in the service of 45 mins+. More important is having the buses spaced evenly between each other but that would require more flexible timetabling & rostering.
Admin:Language edited

[Geo101]

It's interesting that tracking app developers can access data freely. UOW obviously hasn't been able to do that for the Gong Shuttle. Is that because privates in NSW don't have tracking of their buses?

I'm not sure if this is what Ken Tsang (jxeeno) uses, but he also made a website to track the NBN rollout. As far as I'm aware, he has done these sites using publicly available data.

If you use the following link, it displays public transport vehicle locations, from the data.

(The link below doesn't display busses by default, you need to go down the bottom left and click on the B symbol in the Toggle Map Visibility section)

https://anytrip.com.au/map

[tonyp]

It's interesting that tracking app developers can access data freely. UOW obviously hasn't been able to do that for the Gong Shuttle. Is that because privates in NSW don't have tracking of their buses?

I'm not sure if this is what Ken Tsang (jxeeno) uses, but he also made a website to track the NBN rollout. As far as I'm aware, he has done these sites using publicly available data.

If you use the following link, it displays public transport vehicle locations, from the data.

(The link below doesn't display busses by default, you need to go down the bottom left and click on the B symbol in the Toggle Map Visibility section)

https://anytrip.com.au/map

Yes, I'm already onto that thanks. It provides a great degree of analysis, including the vehicle details.

Of course, it is a pretty savage standard to set measuring timekeeping against a 100% perfection. These official statistics that show e.g. 80 or 90% on-time running are measured by being within 5 minutes (or whatever the standard is) of timetable. That makes a helluva difference. The good news is that overnight buses are pretty much on time by a 100% standard lol !

What does concern me a little bit though is the number of bus and tram services running early. That's not good if you're using the journey planner and miss the connection.

[clapham omnibus]

I think there may be something wrong with Katie's interpretation of the data, as it doesn't match TfNSW's monthly aggregate bus on-time running data (see link below). When you look at the 16 months worth of data it appears Transit Systems is the best performing operator in Sydney. Maybe they can help the 370 and STA!

https://www.transport.nsw.gov.au/data-a ... me-running

[andy_centralcoast]

I think in her presentation she mentioned she didn't like Sydney's definition of "on time" so she used Melbourne's criteria instead for her calculations.

[iamthouth]

She is using on time as being zero to five minutes late, at all stops throughout the trip. The TfNSW measure is start point, between 2 mins early to 6 mins late.