For taxi drivers in New York, finding their next fare is second nature, both a function of being in a busy neighborhood, or positioning their car near a stadium or concert hall just when a big event finishes. Uber and Lyft drivers, on the other hand, rely on algorithmically generated real-time surge alerts that direct them to areas of high demand—and increased fares.
Revmax, a Brooklyn-based startup, believes data can trump instinct, and make even better predictions about where to place vehicles. The company’s nascent platform, a data-driven demand forecasting system, helps route cars using event information and machine learning, with the goal of making vehicle fleets more efficient. Ideally, a car will know where to be before you call for a ride.
“There’s a blatant, obvious need for a routing service,” says the company’s CEO, Jonathan Weekley. “There’s a segment of drivers who think they know the right place to go. When you combine simple statistics and machine learning, you can definitely route a vehicle much more efficiently.”
Part of the URBAN-X accelerator for urban tech based in Greenpoint, the company is already working on a user-interface project for the New York City Taxi & Limousine Commission, according to an agency spokesperson. But they have bigger ambitions, according to Weekley: providing some of the brains that will direct tomorrow’s driverless cars.
The idea for Revmax came from the group’s previous startup, an event aggregation platform called Eventflow. The team had successfully created a means to pull together information on upcoming events, as well as a growing user base. But most people used it to find out about events, and never bought tickets, so the startup began looking for ways to monetize the data. One of the companies they approached was Uber. They thought the ride-hailing service would be interested in event information. Turns out, they were right. Events are one of the leading causes of surges, they learned, and cabs, as well as Uber and Lyft vehicles spend more than half their time without passengers on average. Uber’s interest led the Revmax team to rethink and pivot.
Now the team is creating a platform that can direct vehicle fleets and maximize revenue. First, they dug into the data. After collecting taxi, Uber, and Lyft data from New York City to learn general patterns and traffic, they added additional data sets that can help inform and predict demand, including hotel occupancy, restaurant demand, airport arrivals, event calendars, and train arrivals. Machine learning looks for patterns, and helps inform a platform that predicts where drivers should go to get the best fares. The latest model even adjusts for precipitation and temperature. (Uber and Lyft both didn’t respond when asked about how this model compares to their own fleet management systems.)
So far, the team has been figuring out correlations between different data sets and how demand changes throughout the city at any one given time. When they test their program on a fleet simulator, they’ve consistently demonstrated increased efficiency, raising utilization from roughly 50 percent, the industry average, to 70 percent.
“If you route vehicles that don’t have passengers, you can dramatically increase the overall fleet utilization,” says Weekley.
Right now, Revmax is designing a dashboard for the New York City Taxi & Limousine Commission, so they can visualize their entire fleet. Eventually, Revmax’s platform may be able to help direct cabs to the best places to maximize revenue. But right now, it’s just about measuring and tracking. “The first step is visualization,” Weekley says.
But the biggest market by far may be autonomous vehicles. As it stands, companies such as Uber and Lyft can’t exactly utilize this type of technology to its fullest extent, because the vehicle associated with that service are driven by independent contractors and can’t be ordered around.
“Uber and Lyft are using carrots and sticks, such as surge, to route their fleet,” says Weekley. “But they don’t own the fleet, so they have limited control.”
The ability to suggest the best places to go, while tapping into real-time data, could be useful for both today’s human cab drivers, as well as tomorrow’s driverless taxis. In today’s world, there’s a lot of value in a city being able to direct its own fleet, since it can offer huge savings in terms of gas, while potentially easing congestion. Once autonomous cars hit the road, and fleet owners can centralize control, there’s big money in Big Data, and the ability to direct vehicles to maximize efficiency and revenue.