New York Nerds Sift Citi Bike Data to Solve Availability

Each morning, New York commuters who use the nation’s largest cycle-sharing system face a question that decides whether the workday begins in disappointment or with a smooth glide: Will any Citi Bikes be available?

A cadre of data vigilantes, including as many as 80 software engineers, analysts and urban planners, work in their free time to make the answer yes. Without help from the city Transportation Department, they crunch numbers at meetings held by the nonprofit Code for America, analyzing patterns of ridership for the cobalt two-wheelers sponsored by Citigroup Inc. Their tables are lined with laptops -- and sometimes beer.

Sixty-four percent of users say their biggest complaint is not being able to access or return the 6,000 bikes, according to a survey by the Transportation Alternatives, which advocates more cycling. The rebalancing riddle is the primary challenge for more than 500 bike-share programs from Rio de Janeiro to Changwon, South Korea. The most populous U.S. city presents a complex problem that, like the stock market, has a simple exchange at its heart.

“You’re checking out a bike, you’re checking it in,” said Peter Drier, 37, who attends the meetings and is director of technology for prime brokerage services at Wells Fargo & Co. “Those transactions, from a technological standpoint, are very simple, very Wall Street-like. Buy, sell. Check in, check out.”

The city plays constant catch-up with the 93,000 riders to replenish barren docks and make room for returns.

Ubiquitous Fixtures

On Oct. 30, commuters clustered at the vacant 31st Street and Eighth Avenue station, waiting for rebalancers to pedal in with three bikes affixed to trailers on the backs of their own. The cycles disappeared as soon as they arrived.

Three box trucks and nine smaller vehicles shuffle bikes among the 330 stations south of 59th Street in Manhattan and in parts of Brooklyn. The bicycles have become as ubiquitous as the yellow cabs, throngs of pedestrians and relentless honking that define New York City.

Yet jockeying for a bike as commuters roll en masse toward Midtown is a game of chance.

“They should be able to easily do this statistically,” said Todd Hurtubise, a managing member of the hedge fund Proteus Global Partners LLC, walking from one empty hub outside Pennsylvania Station to another. “It’s complicated, but they know when someone takes a bike. They know where they’re going.”

The 46-year-old said he missed a meeting after spending 40 minutes hunting for an empty dock to return a cycle last month.

Crucial Measure

Riders have pedaled more than 9.8 million miles on Citi Bikes, a network that’s about half the size of London’s but sees more daily trips. Citigroup, based in New York, paid $41 million to sponsor them and Mastercard Inc. runs the payment system. The program is run by NYC Bikeshare, a unit of Portland, Oregon-based Alta Bicycle Share Inc., which shares profits with the city.

Scott Gastel, a Transportation Department spokesman, didn’t return telephone calls and e-mails seeking comment on how Citi Bike determines where and when to move cycles or what it is doing to improve rebalancing.

Seth Solomonow, another spokesman, said in an e-mail that “rebalancing is the single biggest daily challenge in the Citi Bike system and these challenges are directly correlated to the success of the system itself.”

Data Scraps

The geek militia working to enable that success meets in rooms at New York University and New Work City, a shared office space in Chinatown. Code for America receives funding from companies including, the philanthropic arm of the search-engine giant, and AT&T Inc. It runs so-called brigades of programmers that have, among other endeavors, enabled people to track vacant properties in Detroit and reunite with lost pets at the animal shelter in Austin, Texas.

In New York, developers with a penchant for bicycling have built more than two dozen maps, automatic graphs and apps for Citi Bike in the past five months using the scant data available. They say they can get only so far without more information. New York doesn’t release information about where and when trips begin and end, unlike systems in Washington and Boston.

“We are told that the data will be shared when it is available,” said Noel Hidalgo, Code for America’s New York City program manager. “Therefore, is the data unavailable? Are there problems within the system that prevent this from happening? If so, that is really scary.”

Extending Visibility

Citi Bike officials have ventured to some hack nights, said Anthony Townsend, a senior research fellow at New York University’s Rudin Center for Transportation Policy and Management, which offers classrooms for the meetups.

“If Citi Bike were to kind of, you know, lift the veil on some of these internal operations, these civic hackers can really help them kind of get a handle on rebalancing,” Townsend said.

University of Chicago researchers partnered with that city to study Divvy, a 4,000-bike network that began there in June. Technology to predict as much as two hours of cycle and dock availability is in the testing phase, said Juan-Pablo Velez, a spokesman for the Data Science for Social Good fellowship program.

“All those stations actually have sensors on them that are broadcasting every minute like a heartbeat,” said Velez, 26. That heartbeat could be irregular in New York, muddling the rebalancing equation.

Multiple Modes

If a station is empty, commuters might walk or hop on the subway instead of waiting until an available cycle emerges from the traffic, said Yang Feng, an assistant professor of statistics at Columbia University.

“They will just leave and you don’t know how many people are doing this,” he said. “You lose track of how many people are waiting there. Only some people are going to wait.”

He said the city could overstock stations and see how many bikes clear out to get a true gauge of demand.

Citi Bike plans to expand into upper Manhattan, Brooklyn and Queens to create a 10,000-bike network and 600 stations. The Transportation Department expects to release a timeline for rolling out the next 1,000 cycles by year-end.

The broadened map could organically rebalance bikes.

“Any activity that increases the proportion of people who aren’t doing the simple twice-a-day commute into the commercial centres will act to counter their effect,” said Oliver O’Brien, a research associate who studies more than 100 bikeshare networks worldwide at the University College of London’s Centre for Advanced Spatial Analysis.

It’s not surprising that Citi Bike has hiccups after just five months, said Hidalgo for Code for America.

“The community that I have been connected to loves to ride bikes because it’s the most efficient mechanism to get around New York City,” Hidalgo said. “Nothing is perfect from the get-go and we recognize that, but that’s why we’re here as concerned citizens.”