Nike+ is a revolutionary product in personal fitness. In its five years, its motivated millions of people to get off the couch and go for a run. At the completion of their run, the user is given a handful of useful metrics (time, pace, and route) in an effort to improve their next run.
What it doesn't offer, however, is a holistic representation of the data. What does it look like when an entire city goes running? And how can we use that data to improve the experience of runners in different cities?
With 1,000 runs of Nike+ data, I set out to do an audit of running in New York City.
After countless hours cleaning the data in Google Refine, and even more hours in Processing, I arrived at one of the simplest visualizations, location.
Though this rendering offers very little information about individual runs and the data within them, I love the story that it does tell about location. The GPS data draws its own map of New York City, from the shape of the Manhattan landmass down to its individual streets.
Popularity of Routes
Next, I wanted to explore how popular these different routes were. I achieved this by creating a simple heat map of all the runs, examining how many runners shared the same route over the course of this data set.
New York City - Popularity
Not surprisingly, Central Park and the trails along the edge of Manhattan emerge as the most popular, as well as the bridges between Manhattan & Brooklyn. Downtown Brooklyn, especially along the promenade, also sees a great deal of traffic. Interestingly, it appears that more runners in Central Park tend to come from the Upper East Side, with far less entering the park from the Upper West Side.
Central Park- Popularity
Bridges - Popularity
Direction of Travel
In track & field, runners always circle the track in a counter clockwise direction. But out in the world, our routes and directions are far less prescribed.
I was curious to see if there were certain areas where runners all ran similar directions, regardless of no set of rules. If a runner’s position changes towards the south, their path is drawn yellow. If a runner is heading north, their path is blue.
Central Park - Direction of Travel
In looking at the Central Park Loops, almost all runners choose to run in a counter-clockwise direction, following track protocol on all loops, regardless of size. We can also see a balance of entrances and exits to most parts of the park, except for one spot on 59th that seems to cater exclusively to runners entering the park.
Bridges - Direction of Travel
The Brooklyn Bridge is being used almost exclusively to funnel runners out of Brooklyn, while the Manhattan Bridge is used to cross from Manhattan to Brooklyn. It appears these bridges were named for the runners leaving the borough.
The Williamsburg Bridge, as opposed to its counterparts with their paths down the middle, has two distinct running paths on each side of the bridge. Here we can see, interestingly, that runners seem to favor the side of the bridge going against traffic.
Time of Day
So far, all of the runs have been examined simultaneously. But in order to look at what time of day the runs take place, I elected to create timelapse videos. All runs, regardless if they took place on September 5th or December 12th, are displayed in the same day, mapped to the exact second that the GPS data was recorded.
New York City - Time of Day
Central Park - Time of Day
Next, I wanted to explore the distances that New Yorkers run and how it might be correlated to the different parts of the city that these runs are taking place.
New York City - Distance
Not surprisingly, longer runs tended to be more prevalent in areas that runners could get to stretches of uninterrupted running trails, such as Central Park, the bridges, and the West Side Highway. Landlocked areas where trails are replaced by streets tended to see much shorter runs.
I was surprised to see that most of the runs originating from the Upper East Side and Upper West Side were short runs, given their proximity to Central Park. It appears that people who begin their runs in Central Park tend to go for longer runs, while those who start outside of Central Park and run into it tend to go for shorter runs.
Similar to distance, I wanted to explore what different parts of the city did to a runner's pace. I went in with the assumption that somebody running in Central Park would be running a faster pace than someone in Midtown, due to the minimal obstacles like stoplights, traffic, construction, and other hazards that runners in the middle of the city face.
New York City - Pace
Not surprisingly, the fastest runs took place in Central Park, and along the southern edges Manhattan.
The Brooklyn Bridge, though overly crowded with tourists and bikes, still allowed for fast runs. The Manhattan Bridge was a little slower, and the Williamsburg Bridge, which in my personal opinion seems to go uphill both directions, had drastically slower runs.
The previous exploration compared the overall pace of a runner against all of their New York counterparts. It leads to some interesting insights, but it doesn't account for the fact that not all runners run at the same pace, and as a result, the total pace might lie more in the individual runner, than in the city and its clear/crowded running trails.
So, to go a step further, I attempted to examine an individual runners pace at any given moment compared to the total pace they would finish their run with. If they slipped more than 20% below their total pace, that segment of their run was drawn in red. If they were 20% above their average pace, the route was drawn in green.
New York City - IndividualPace
Now we can clearly see the parts of the city that post a problem to New York runners. Midtown emerges as a terrible area for running, as almost everyone's pace slows as they hit countless stoplights and crowds.
Again, the Williamsburg and Manhattan bridges lead to slower paces in comparison to the Brooklyn Bridge, which allows for some brief glimpses of faster than average paces. My hypothesis is that the Brooklyn Bridge has a less severe slope than its neighbors, which allows runners to gain a little momentum on the downhill while not being hindered too much on the uphill.
Finally, its worth noting that not all of Central Park leads to faster times. The northwest corner, where I believe there is a large hill, leads to much slower times. There's also a long stretch on the east side of the park where runners seem to slow.
Finally, I wanted to examine the points of a run where runners come to an absolute standstill, most likely due to traffic signals. If a runner's GPS position didn't change after 10 seconds, a red circle was drawn. The longer they were there, the more intense the circle.
New York City - Interruptions
I was surprised to see how widespread these stoppages were. I anticipated a great deal of them to be outside of the park runners were slowed by stoplights on the way to the park, but was surprised to see how many surfaced within the park.
Runners in Brooklyn seemed to hit less delays than their counterparts in the interior of Manhattan. Those running on the outside of Manhattan seemed to hit less, though almost every pier on the West Side Highway seems to have a runner who ran out to the end and took a break.
The above audit of New York City running took place outside of class, after the semester had ended. Initially, Nick's assignment was only two weeks long, which in grad school time, means one weekend of nonstop work to complete. Below, you will find my initial study, and what I presented to the class and our guest panel after these two weeks. In hindsight, I feel that these videos, though nice looking, were a bit convoluted in the information presented and what a viewer could extract from them. Additionally, I came to realize that though time of day was interesting, it was just one of many stories that could be told, as demonstrated above. Finally, in my rush, I realized that I made small errors in my data cleaning, such as not accounting for daylight savings time, that I have since corrected.
Having said that, I'm extremely proud of what I accomplished in a short time, especially in regards to programming in Processing.
I was interested in visualizing not only where these runs were taking place, but also, at what time of day. I decided to create a timelapse video, displaying each individual route at the time of day it took place. Moreover, I elected to use a map of New York City, but rather, let the running data generate a map of New York City as the runners move throughout the city. Paths, streets, and bridges all start to emerge as the day goes on and more runners move through the city.
Version 1 :: Green to Red
Version 2 :: Black to Red
Professor :: Nicholas Felton
This project is inspired by Aaron Koblin's Flight Patterns project.
Featured :: Wired, Fast Company, The Fox is Black, Information Aesthetics
Please note, this project was completed separate from my summer internship with Nike.