This week we should learn about the patronage results for September and with this post I want to explore whether Auckland Transport are delivering the results to the public in the best way that they can.
Currently we get patronage results a couple of reports that go to the AT board each month. There is the Public Transport Monthly Patronage Report, the Monthly Transport Indicators, the Statistics Report and even some details about HOP usage in the Chief Executive’s Report. Each offers the same high level information but there are variations between them. I tend to use the Statistics Report as that generally has the more detailed information than the other reports. The fact there are multiple reports to begin with is odd and at the very least the Public Transport Monthly Patronage Report, the Statistics Report and the HOP reporting from the CEO’s report should be combined together in a single report.
Other than the number of them, there are a couple of other issues I have with the reports. The primary one is that they are only available as a PDF report. That means each month I have to go through the report and pull out all of the details manually if I want to keep track of them (which I do and I know some others do too). This opens up the chance of data entry errors with the information or incorrect numbers if a figure is revised which happens from time to time and happened recently with the ferries. You also have to know that the patronage results are included in the board reports and where those reports are buried on the AT website. Other issues relate to what information is available compared with what other cities provide.
So with that in mind here are some examples of what some similar organisations provide to the public.
Wellington
Greater Wellington Regional Council (GWRC) through Metlink recently improved the level of information they provide and importantly do so in an easy format for anyone wanting to look at it. They provide a range of graphs showing the monthly results for the current financial year or the annual results as far back as 1999/2000 and most of the data is available in a spreadsheet that can be downloaded. The data provided includes many of the same types of areas that AT provide but there are some important additions. In particular
- Annual peak and off-peak patronage – this shows how much patronage occurred during the peak and off peak and in the spreadsheet is also available by mode.
- Annual passenger kilometres by year and mode – This shows how far people have actually travelled on each PT mode which is useful for seeing how commuting trends are changing. As an example on average bus trips are getting longer while rail trips are getting shorter.
The one downside to how GWRC produce their PT information is there is no context able to be given, for example patronage that is impacted by special events or holidays etc. The results are updated approximately 1-2 months after they occur.
Perth
The Public Transport Authority runs PT in the Perth through their Transperth brand. The authority provides monthly and annual patronage information via an online interactive table by mode and for trains by line. It’s not clear how frequently the information is updated however as the image below shows, it’s not as frequent as Auckland or Wellington. There are no graphs or any contextual information however. There’s also no information on other metrics
Portland
PT in Portland is run by TriMet and they provide a number of ways for the public to get patronage information. Firstly there is a Performance Dashboard which shows graphs about the average weekly boardings per month (instead of total patronage) but most interestingly they also provide financial information including the average cost per trip and revenue. Reporting revenue monthly is particularly interesting as in most cities you have to delve through dull Annual Reports to find the information hidden in the financials – although even this isn’t possible with Auckland Transport as it isn’t specified in their annual report.
In addition to the Performance Dashboard also publish monthly reports which includes all of the figures from the dashboard plus a few others and to top it off the data is also available back to mid-2008 in one file.
One of the more interesting aspects about all of the TriMet data is how they break the bus data down by whether the bus is a frequent route (at least 15 minutes all day) or a local connector route. In Portland frequent buses carry over 50% of all bus patronage. As Auckland Transport roll out frequent buses as part of the new network here I hope they differentiate between the frequent and non-frequent services too.
It would be great if AT could also provide operating cost information regularly
San Francisco (BART)
San Francisco is unusual in that the Bay Area Rapid Transit (BART) system is run completely separately from the rest of the PT services in the Bay area. The patronage information BART release doesn’t show the total number of trips, instead it shows the average daily ridership for a weekday, Saturday or Sunday. One of the advantages of using an average weekday result is it more easily accounts for the variations of the calendar and is something Auckland Transport have recently started doing. Instead of just showing the overall result the monthly data goes a step further by using an Entry/Exit Matrix which shows the average daily ridership from each station to each other station on the network. The image below is from last month and as an example it shows that on average for a weekday 852 people catch a train from El Cerrito Plaza (EP) to Berkeley (BK). This is a level of detail is likely to only be practical to provide for a rapid transit system and something I think AT should definitely do for both the rail network and the Northern Busway.
In addition to the level of detail the files are updated quickly and are usually available by the 5th of the next month (compared to almost one month later in Auckland). Lastly one extra feature is that a spreadsheet is available with the annual patronage information back to when the system opened in 1973
Conclusion
So what could AT learn from these cities to improve how it provides information on patronage to the public.
- At the very least:
- consolidate the various reports into a single report that contains all the relevant information
- a page on the AT website with links to each of the monthly patronage reports to the board.
- Should have:
- A page on the AT website with some graphs explaining the key PT results
- Provide a downloadable file with historical patronage results
- Would be nice to have:
- An Entry/Exit Matrix for the Rapid Transit network (rail and busway)
- Data updated automatically earlier in the month
- Operating Cost and Revenue information
- Would be ideal but won’t hold my breath for:
- An Entry/Exit Matrix for the entire PT network that the data wizards out there can use to create new insights into our system.
Is there anything else you would like to see?
Processing...
Reliability and punctuality done similarly could also be useful – especially now that the bus numbers are no longer fraudulent.
I think PTOM requires the published ranking of unit performance, which is great.
Cash versus HOP by route/time of purchase. This would allow a comparison of actual charges against fare box recovery, along with some other metrics around trip length, percentage paying cash, time of day cash is paid, areas where cash is the dominant payment method and probably some others I can’t think of.
This data set would be at the trip level, so that you could look at variance between day of week and should contain the route number, which would allow the inclusion/exclusion of special event information.
Matt, you can convert the PDFs to Word or Excel, although having done that with the Statistics Report I see that the graphs are all embedded as pictures, although the tables can be extracted either via Word or directly into Excel. The Word version is a lot more readable, so copying and pasting tables from Word to Excel is tidier. Unfortunately the formulas are lost (ie everything appears as values), which may be how they were loaded. Still, it might save you a bit of time (unless you already do this).
I wholeheartedly support Matt’s recommendations here and I also recommend that AT work out a way to make the raw HopCard data freely available online and updated daily. This should be at the lowest level of detail possible without naming individual users. Basically their raw transaction data. This would get hacked about analysed and modelled. Given the amount of free advice AT get from this site I’m sure some data geeks would not only work out ways to prepare the reports Matt recommends but also do all sorts of other reporting that AT haven’t thought about. A crowd sourced solution for free.
I was thinking about the frequency of uploads and daily was the answer I came up with, although I’d expect it to be marked provisional until the end of the month processing and normalisation processes had been completed.
I’d be against the raw feed, it would need to be aggregated in some ways to remove personally identifiable elements, so you wouldn’t be able to track an individual through the publicly available data.
It would be great to see the graphs and data presented in a 2014 and beyond format and approach. The graphs and tables look like my first attempts in Excel from 1995 which was definitely not flash. Maybe a data consultant could quickly work up some visually presentable and meaningful ways to tell the story i.e. what is up, what is down, changes in modal usage, customer segments and patronage behaviour across days and times of the day.
Having a dashboard approach would be a good start for a quick snapshot/executive summary with further details in below.
As much I know some of the data geeks on this blog would love to get their hands on trip detail and HOP transactional data I suspect that a lot of this is commercially sensitive until PTOM is completed. Would be great, but would start a level of questions that AT would probably be uncomfortable fronting. Would be refreshing to get a level of ‘transparency’ but this would probably need a fine line to be walked.