News › Stories › Archives › 2019 › May › Predicting Demand Improves Ridesharing Experience for Drivers, Passengers May 10, 2019 Predicting Demand Improves Ridesharing Experience for Drivers, Passengers
By Jacob Williamson-Rea jacobw2(through)cmu.edu
Media Inquiries Sherry Stokes
- College of Engineering
Ridesharing pricing models are based on supply and demand. When demand spikes, prices surge, and when demand is low, prices are based on a flat base rate. Drivers enjoy higher profits in popular areas with surging prices, but current pricing models function in real-time, meaning that drivers only see busy areas after the areas are already busy.
As a result, both passengers and drivers lose out. Passengers wait longer and potentially pay more, while drivers try to catch up with changing rider demands. Additionally, when drivers hit the roads without a plan, they end up coasting and looking for riders, which not only wastes time and fuel, but also worsens traffic congestion.
To improve ridesharing services, Sean Qian, an assistant professor in Carnegie Mellon University's Department of Civil and Environmental Engineering and director of the Mobility Data Analytics Center, teamed up with Gridwise, a Pittsburgh startup company founded in 2016 that aims to improve on-demand ridesharing. The group received funding from the Pennsylvania Infrastructure Technology Alliance (PITA) to develop an advanced predictive model that makes ridesharing platforms more efficient. Now the Gridwise mobile application allows drivers to track real-time rider demand, as well as plan ahead of time for upcoming surges in driver demand.
Several years ago, Ryan Green, now the CEO and cofounder of Gridwise, drove for rideshare companies like Uber and Lyft. He grew tired of chasing surges and founded Gridwise in 2016 to help his fellow drivers.
Civil and Environmental Engineering Assistant Professor Sean Qian discusses using real-time data to predict future traffic volumes and reduce congestion and emissions.
"Gridwise originates from the perspective of the ridesharing drivers," Green said. "We saw the problems that existed for drivers, so we explored ways to help drivers combat these every day challenges." Green and the Gridwise team joined up with Qian, which allowed them to utilize extensive data sets to better understand what makes passengers and drivers a good match.
The Mobility Data Analytics Center provided data about transportation including parking, traffic incidents, and bus ridership, which serves as a foundation for the model. Qian and Matthew Battifarano, a graduate student researcher, combined this foundation with social media data, ridesharing data and more provided by Gridwise, and developed a model that can now predict when and where ridesharing demand will spike up to two hours in advance.
"We use all kinds of information and data, from weather conditions to traffic incidents," Qian said. "This way, if a driver sees a possible surge in an area related to rain or a concert, they can make a more efficient decision."
Green says about 15% of Pittsburgh's ridesharing drivers used the Gridwise application when the PITA grant was awarded, but now the majority of the Pittsburgh's ridesharing drivers take advantage of it. The app spikes drivers' profits by up to 40%, and riders enjoy shorter wait times as well.
Gridwise has expanded into other cities, too, which means Pittsburgh problem-solving isn't just helping Pittsburgh drivers—it's improving ridesharing experiences and traffic congestion across the United States, making the country's transportation grid more efficient.
"This PITA-funded project provided us with the platform to solve a real-world problem here in Pittsburgh," Qian said. "We've effectively improved the mismatching that occurs between drivers and passengers, which means better service for passengers, increased revenue opportunities for drivers, and overall better transportation infrastructure performance."