Passenger flow data - ticketed journeys between stations

[whole diary]

[grahame]

Explanation:

I have registered the Coffee Shop for data feeds from Raildata - big THANK YOU to the member who pointed me in the direction of that (and also explained where a couple of other sites get their data).  Early experiment times - 86 tables of data, many require payment for access, but many others are under an Open Government Data License.  The data is big - I have downloaded passenger flows from station to station for 4 years - 2018/19 to 2021/22 (the latest available so far - 2022/23 expected this autumn) and it's over 5 million record with a total file size of over 400 Mbytes.  And then the problem is how to access that and make it useful on a live server within the resource we have!   An enjoyable few hours over the weekend and I have filtered the data and imported it into a database - searches for data are still substantial requests on data of that magnitude, but I have been able to make it available.   I am monitoring usage so as to impose limits if I have to.

Usage:

Click on the map at the top left of a forum page



That will take you to a bigger map (it may not be exactly this one - you can set preferences for this)



Click on the station of interest and you'll go to a page for that station



And at the top of the middle column click on "Passenger flows" to give you



You can then click on individual stations in the list to see the flows from there, and / or replace the 4 digit station code number or three letter code in the URL with any other station of interest.

I am aware that there are quirks in the ticketing data but never the less it's amazing what I have already learned in my experimentation - for example confirming the numbers and importance of flows from west of Didcot to Oxford.

Edit to add - intermediated URLs in the above sequence:
http://new.passenger.chat/better/map.html
http://new.passenger.chat/better/map.html?stn=MDN
http://www.passenger.chat/flows.html?stn=MDN

[grahame]

A developing application - now, once you get the flows up for a station (initially sorted by their rank in 2018/19), you can click on a column header to see them ranked by another year. It seems sensible to provide an option to see how they are doing in the latest data, as well as preCovid.

[Red Squirrel]

Fascinating stuff, grahame - thank you! Who would have thought that the fourth most popular destination from Montpelier would be... Paddington!

[grahame]

Fascinating stuff, grahame - thank you! Who would have thought that the fourth most popular destination from Montpelier would be... Paddington!

It WAS pre-Covid ... but no longer. Dropped to 11th in 2020/21 then 9th in 2021/22.   I suspect that's mostly because of the massive growth in local traffic on "The Beach" (or is that partly a growth in fare collection?) rather than than an asymmetric loss of London traffic.   Complex stuff!

[PhilWakely]

I guess figures will now be distorted with the growth in split ticketing. It would be interesting to see figures of journeys into the likes of Didcot, Pewsey and Castle Cary.

[grahame]

I guess figures will now be distorted with the growth in split ticketing. It would be interesting to see figures of journeys into the likes of Didcot, Pewsey and Castle Cary.

It will be interesting to see the next year of figures.    The ORR^» (Office of Rail and Road formerly Office of Rail Regulation - about) methodology report for the 2022/23 year (mirrored (here)) tells us

In the April 2022 to March 2023 dataset the following methodological improvements have been implemented:

* An adjustment to account for split ticketing, where a passenger completes one journey using two or more tickets.

and I am pretty sure the ORR figures are the base / same data as the new tables we have.

[Red Squirrel]

Ah, yes, so the 'ranking' in the left-hand column is based on the 2018/9 figures...

I'm sure the Severn Beach line stats will have been affected by more correct assignment of origin and destination stations recently. Not so long ago everything between BRI^» (Bristol Temple Meads - next trains) and CFN^▸ (Clifton Down station) was ticketed to the end points, so any stats for LWH, SRD, MTP and RDA^▸ (Regional Development Agency) could only be rough estimates.

Incidentally I note that the querystring variable (CRS code) is case sensitive, so

http://www.firstgreatwestern.info/flows.html?stn=CИM

gets you Cheltenham Spa (if you use a conventional N), whilst

http://www.firstgreatwestern.info/flows.html?stn=cnm

isn't recognised and so, quite rightly, defaults to Melksham.

[grahame]

http://www.firstgreatwestern.info/flows.html?stn=cnm

isn't recognised and so, quite rightly, defaults to Melksham.

That was an unplanned feature. It now accepts 3 lower case letter as CRS code.

[eightonedee]

Out of personal interest - how does it account for season tickets?

I noted with interest that the 2020-21 figure (1st lockdown year) for Goring & Streatley to Guildford was only 127, whereas the previous two years it was 1974 and 1264. In those years I would have been doing about 200 journeys on that route on annual season tickets. Can I see the effect of lockdown on my commuting habits reflected in these figures?

[grahame]

Out of personal interest - how does it account for season tickets?

From the document quoted above:

Season ticket journey factors – Ticket sales in LENNON^▸ (Latest Earnings Nationally Networked Over-Night ) are converted into an estimate of the number of journeys made by applying a series of ticket type journey factors. Therefore station usage estimates are based on an assumed number of journeys made based on the ticket type sold. The journeys factors used for the main season tickets are as follows:

Season ticket validity - Journey factor
Weekly         10.3
Monthly     45
3 monthly     135
6 monthly     270
Annual     480

And more with specific issues in London, Southend, and Digby and Sowton seasons!

[grahame]

I noted with interest that the 2020-21 figure (1st lockdown year) for Goring & Streatley to Guildford was only 127, whereas the previous two years it was 1974 and 1264. In those years I would have been doing about 200 journeys on that route on annual season tickets. Can I see the effect of lockdown on my commuting habits reflected in these figures?

Yes - at the lower end of the data you can spot individual's effects ...
http://www.firstgreatwestern.info/flows.html?stn=4228&limit=50&sortby=1

I speculate that someone had a regular weekly commute from Knucklas to London (Euston) preCovid but is now going into the office one a month, and that someone else started commuting to Hereford in the run up to Covid and has stuck to the trains through thick and thin.

If you change "limit=50" to "limit=1000" for any station you can see a huge number of tiny flows even for the busiest of stations.

[Mark A]

Then, from Will Deakin, this: https://github.com/anisotropi4/kingfisher/blob/main/station.md

Article: https://www.bloomberg.com/news/features/2024-02-08/british-rail-train-ticket-data-maps-show-how-a-nation-travels-before-hs2

Mark

[grahame]

Data for 2022-23 published, and added to our database.

Example for SWI^» (Swindon - next trains) :
http://www.passenger.chat/flows.html?stn=SWI

[grahame]

At the base of each flow data page, I have added a section showing the (changing) percentage to / from each destination.  They can be navigated to via the disruption map  - sample URL http://www.firstgreatwestern.info/flows.html?stn=BOA&limit=50&sortby=4 if you want to play directly.

Here's an example -


These flow reports make fascinating reading, and food for thought.  I notice so much the dominance of direct service destinations - does that indicate that the service broadly goes to the right places, or that there is a suppressed / constrained flow to elsewhere because of the need to change trains and perhaps wait?

The appearance of both Bristol and Paddington in the top ten from Melksham in spite of the change suggests that there are significant opportunities to expand that market - it is not an easy journey with waits that add up to 35% in time at interchanges, and a need for knowledge at Bristol on the return trip as to whether to take the service via Trowbridge or via Chippenham - both in terms of the timetable and knowing ahead of any disruption which would change the choice.   Melksham < - > Bristol is highlighted as a flow and a connection that needs to be carefully considered - improved connections (a couple of key trains shifted by as little as a minute will save long waits!) better information and publlicity and an improved interchange at Trowbridge where the lack of slopes or a lift makes the change very difficult for people with heavy luggage, with small children and the equipment involved, or those in a wheelchair.

[jamestheredengine]

I find it slightly surprising that the 38th busiest destination from Neath is Birmingham Moor Street. Yes, it's possible to go there on a Birmingham Stations ticket, but you'd basically have to be mad to try it.