RFID Data Reveals London's Polycentric Pattern of Commuting
Data from London's commuters reveals travel patterns in unprecedented detail. It also raises privacy fears.
kfc 01/29/2010
- 3 Comments
Since 2003, some 10 million RFID cards have been issued to commuters using the London transport network. The cards are precharged with credit which is then used to pay for journeys on buses and trains.
For scientists studying the behaviour of commuters, the system has generated a firehose of data. Camille Roth at the Institut des Systemes Complexes in Paris and a few buddies have got their hands on the data for the 11.2 million trips that took place during the week of 31 March to 6 April 2008. This data included the start and finishing point of each journey as well as the journey time.
The data gives one of the most detailed insights into commuting patterns ever assembled. It has long been known that London has three main centres for commuting: the West End, the City and the Midtown area between them. However, nobody has been able to tease apart the structure of commuting in any higher resolution.
Until now. The new data shows for the first time that that there are numerous smaller centres as well. These can be ranked according to their total inflow each day. Just behind the big three are centres such as the Docklands area in East London, the area around Parliament and the area to the south west of this near Victoria and Green Park, which the authors call the Government area.
At even higher resolution, it is possible to see centres of inflow around the museum area of South Kensington, the Northern business area around Camden Town and so on.
That's not entirely unexpected but it is a step beyond the standard theoretical description of city commuting.
The reason it is important, of course, is that it gives city planners a starting point from which they can model the effects on th transport infrastructure of new urban projects.
The big fear, of course, centres on privacy issues. It isn't clear to what extent the data is anonymised before it is released. Could it be possible that some suitably clever individual with an electoral register and a few business directories could begin to link individual journeys with specific individuals?
After the Netflix and AOL scandals, only the brave (or perhaps stupid) would rule it out.
Ref: arxiv.org/abs/1001.4915: Commuting In A Polycentric City
On Twitter
Become a Fan on Facebook
Subscribe to the Feed
colinnwn
88 Comments
Not very likely
"Could it be possible that some suitably clever individual with an electoral register and a few business directories could begin to link individual journeys with specific individuals?
After the Netflix and AOL scandals, only the brave (or perhaps stupid) would rule it out."
Is it possible, probably. But is it likely, no. The Netflix data contained a much richer set of data about each "anonymized" person, which was able to be correlated with publicly known data. Each of the many movie ratings provided unique bits of information.
From this quick blurb, it sounds like the Underground data contained only a rider number, time/location on, and time/location off. Additionally the limited stops condenses a lot of potential unique address information into large groups of riders going to many addresses in the walkable area. You'd have to use a lot of independent observations, beyond home addresses, electoral roles, and business listings, to tease a specific person out.
That is practical to do if you have an interest in a specific person, but not so practical in aggregate.
Reply