Data-mining Oystercard to detect travel patterns in London

 by Martin Belam, 25 September 2003

I wrote before about my failed attempts to purchase an Oystercard online due to the "shonky" e-commerce bit of the website (©2lmc) - but I eventually resigned myself to using Internet Explorer and being outside a corporate network firewall to do the transaction. So I am now happily in possesion of a smartcard tracking my every movement around the London transport network. Frankly between that and my mobile phone I assume that someone, somewhere, has got me pinned down to a specific location at any given moment of the day. David Blunkett will have to try harder (and apparently charge me £40 for the dubious privilege).

But using the Oystercard made me realise that if TfL can switch most regular London travellers onto this format they have the most amazing set of data to investigate. There is potentially a lot of data available for analysis which has never been seen accurately before, linking particular patterns of use between the underground, mainline trains, and buses.

Some of the thoughts that occurred to me were:

Monitoring peak flow at stations - if you know, because you can track individuals from arrival to destination, that a five minute delay to an individual train departure increases the usage of the train by a disproportionate amount, can you regularly delay departure to cram more people on to that train, or shift the departure earlier to better distribute capacity?

Adjusting timetables - if you can track that n number of people leave station o at pp:pp during rush hour and then board bus q or train r, can you adjust the timetable to connect better? and if you see that more people from route x then use route y rather than route z, could you prioritise not just the timetable of route y, but also the place of departure in order to minimise people's transfer time? Or do you arrange the station in such a way as to ensure that the maximum number of people walk past the retail outlets and advertising in order to increase their commercial value?

Adjusting the flow through stations - if you can prove that a greater percentage of travellers change from line x to line y at a station like Kings Cross, can you adjust the positioning of the ticket barriers and the rush hour one-way flow to make that journey the quickest and easiest?

Delivering information - if you can show that a significant number of people using one set of ticket barriers move on to the same service, can you provide them with timely information more quickly - for example at London Bridge or Liverpool Street ensure that the displays give the most used services priority in the visible information available from the appropriate entrance / exit?

...and on a lighter note, can you use the technology to invent new words, like the station announcer at Finsbury Park overground on Saturday afternoon, who let us know that due to faulty visual displays we should listen out for announcements about "your destinating trains"

Keep up to date on my new blog