Every time a commuter in London swipes an Oyster card when boarding a bus or exiting the Tube, that transaction is recorded as one of millions of points of data on a given day tracking the choreography of the city. In aggregate, this dataset is mind-bogglingly massive, cataloging the commutes and minor movements of hundreds of thousands of people back and forth across town.
How do you make some useful sense of the patterns in a dataset that large? For one thing, you visualize it. The below animation comes from transportation planner Jay Gordon, who has turned 16 million transactions on London public transit from a typical Wednesday in 2011 into a one-minute time-lapse of the city:
That animation is built on private Oyster data from Transport for London shared with researchers at MIT (this data isn’t public in part because it contains potentially sensitive information about the commuting patterns of individual people). As part of his master’s thesis at MIT, Gordon devised a method of inferring travel patterns for commuters using the Oyster data (bus commuters, for instance, only swipe their cards when they board a bus, not when they hop off, leaving academics to infer their full commuting patterns with algorithms).
In the above animation, the green dots represent passengers riding either the train or bus (the brighter the dot, the more passengers). The blue dots represent where those passengers were located either before their first ride of the day or after their last – likely a glimpse of their home location. And the red dots show cardholders in between transit trips – in other words, people at work, or in the mall, or using other transportation like taxis.
Over the course of a day, starting at 3 a.m., we can watch London get out of bed, travel from the outer boroughs to the city’s core, run around at lunchtime, head home again around 5 p.m. and then turn in for the night at those static blue locations. The animation includes some rough approximations. Passengers in the London Underground, for instance, appear to move in a direct line from their origin to their destination, not along the precise route of the underground rails. The animation also excludes passengers using paper tickets (a small minority in the system). But Gordon’s picture does account for what we know about how long it takes each of these people to make a transit trip (using either Oyster data, or GPS bus tracking).
Gordon built the animation with your basic red-green-blue color spectrum to allow for the overlap of all those glowing dots. In the middle of the day, for instance, the center of town appears in yellow, as most people there are either in transit (green) or what we assume to be at the office (red). “Pure white,” Gordon explains, “is everything – everything is maxed out, tons of people are at home, and at work.” And commuting between the two.
The travel patterns inadvertently reflect land use in the city. The knots of activity downtown during the day reveal London’s commercial core. Farther out from the center of town, some arteries appear as red while others look blue; the red ones are likely commercial corridors with heavy bus travel. At higher resolutions, a few black spots in the city even capture public parks and the River Thames, where no one is commuting, working or sleeping.
Gordon’s goal wasn’t merely to create a tantalizing animation. Visualizations like this could actually be used by transportation planners. “A planner comes in in the morning,” Gordon envisions, “has a cup of coffee, watches this one-minute animation from yesterday and looks for patterns. If something looks weird, they can say ‘what happened yesterday over by King's Cross Underground?’”
Maybe a train broke down, or a bus got stalled in traffic. A planner could then zoom in on the visualization at higher resolution, or pull other related data reports. The initial process obviously requires a computer. “But it still relies on a planner’s expertise,” Gordon adds, “to notice something that looks funny.”
None of this would be possible before the advent of magnetic transit cards. Now they allow transit systems to track not just individual rides, but the ridership patterns of individual people across multiple trips in time. Without the Oyster card, we could still see a flurry of trips in the morning, then more in the afternoon. But we wouldn’t be able to connect them to each other or a coherent whole.
“Even if you had a powerful computer 10 years ago, the only way to collect this data then was through surveys, which people still do,” Gordon says. “It’s really extensive, and you get a small sample. The strength of this is you get a massive sample, every day of the week.”
That one animation of millions of data points represents just one day in time. London tomorrow – or on Saturday, or during a winter storm – could look entirely different.