Making sure the trains run on time, it turns out, is not as easy as it may seem.
In fact, with ridership on public transportation at an all-time high, more and more transit authorities around the country are turning to data analytics to improve their services, said Tom Kowalski, president of Urban Transportation Associates (UTA), a transportation consulting firm.
"We're in the transition from a very old data collection method to a more modern data collection method," Kowalski said.
Until recently, he said, many transit authorities hadn't changed the way they devise bus routes or determine the number of trains to run in 20 or 30 years. They often used anecdotal evidence gathered by bus drivers and train conductors, for example, to assess ridership patterns.
But the sheer number of riders is quickly making the manual methods of collecting and analyzing ridership ineffective. According to the American Public Transit Association (APTA), public transportation ridership has surged by 38% since 1995. In 2008, that translated to 10.7 billion trips on public transit, APTA says.
To help transit authorities get a handle on ridership trends, UTA has developed a system it calls Automatic Passenger Counting. UTA installs sensors on trains and buses that collect data on the number of passengers getting on and off, as well as the number of stops the vehicle makes and runtimes, much like a black box in commercial airplanes.
Employees can run a numbers of reports and "what-if" scenarios, based on their level of expertise, to determine, for instance, where a new bus route might be needed, where cutbacks on service because of a lack of usage might make sense, or whether there are too many or not enough buses serving a particular route, Kowalski said.
The Metropolitan Atlanta Rapid Transit Authority (MARTA) has been using UTA's services and SPSS software for more than 10 years to conduct a variety of analyses to assess route productivity on more than 130 bus routes with a daily ridership of 500,000 people, according to Tonya Saxon, transit system planning analyst with the authority. She said MARTA uses the software to generate trip reports to respond to overcrowding complaints and to evaluate ridership at the "stop-level to identify underutilized stops for MARTA's Bus Stop Re-Spacing and Consolidation Project, which will work to reposition and consolidate many bus stops in the system."
Before using UTA and SPSS, MARTA conducted similar analyses manually, Saxon said. "Manual data [collection and analysis] had limitations and was not as effective for use in conducting geo-coding (stop location) and route segment analyses," she said in an email interview. "The reports allow us to assess infrastructure utilization, trip productivity, route segment productivity, and route productivity."
Each transit authority has different levels of user expertise, however, and Kowalski said SPSS, which is in the process of being acquired by IBM, gives him the flexibility to add or remove features and analysis tools to meet end-user capabilities.
At the MBTA in Boston, for example, workers from the Massachusetts Institute of Technology's transportation department assist with some of the ridership analysis. At MARTA, each department has at least one staff member trained on the software that runs script-based reports and taps UTA for assistance as needed, according to Saxon.
"There is a large range of analytic capabilities in the marketplace," Kowalski said, but transit authorities are increasingly hiring business analysts to take better advantage of the ridership data collected.
Despite recent budget cuts that some transit authorities have endured, Kowalski expects -- thanks to an influx of federal stimulus dollars -- more transit authorities to invest in analytic capabilities to improve operations and ultimately drive ridership even higher.
He also said volatile gas prices and pressure from the government and the public for more environmentally friendly transportation options are likely to further increase public transit ridership and, hopefully for UTA and SPSS, demand for better analytics to measure and respond to it.