This article was originally published by Mobility Lab.
TRANSPO TAKEAWAYBy providing – and improving – real-time system data and working with third-party developers, agencies can help riders better understand potential issues with their commutes and make smarter choices.
While Metrorail and Metrobus ridership has fallen in recent years, the transit system is still an essential backbone to the Washington, D.C., region’s transportation options. Keeping riders aware of daily system performance is a legion of concerned programmers, working to make it easy and predictable to ride Metrobus or rail.These developers and Metro enthusiasts came out in force for February’s Mobility Lab-sponsored Transportation Techies event, Metro Hack Night IV, one of the Meetup group’s highest-attended events, and its first in the headquarters of the Washington Metropolitan Area Transit Authority’s downtown headquarters.
WMATA’s transit data
The evening began with a long-awaited announcement from the agency: WMATA will soon be providing the public with real-time train positions, a piece of data that is also of much use to developers. While the agency currently provides real-time arrival predictions, train positions is an essential piece of data that would allow riders to better understand, and plan around, how the system works. In the same vein, WMATA will be hosting several “data day” events to work more closely with and listen to the concerns of these developers who employ the agency’s data.
WMATA Department of Information Technology’s Mary Kaye Vavasour led off the night with a presentation on the agency’s Application Programming Interface, which packages and delivers data like train schedules and arrival predictions to the many apps that utilize that information.
The department has reached a stage in which it is engaging with developers to “foster collaboration on focus and investment” in providing data, as well as “fixing what’s broken” with the data, according to Vavasour. For example, about 89 percent of data requests that come in from third-party apps are for real-time bus and train data. Based on this revelation, the team spent six months specifically improving bus prediction times and real-time data.
Since fixing bus prediction data and engaging more closely with developers several years ago, data requests have doubled each year, with 600 million calls in 2014 and 1.2 billion in 2015. Each “call” represents an app user requesting a piece of data – an arrival time, a stop location – about the Metro system, a clear indicator of the strong demand for reliable Metro service information.
Putting the data to good use
The remaining Transportation Techies’ projects, offered in lightning-round style to squeeze in the long list of presenters, revealed several key themes of how people interact with the transit system on a daily basis. In conjunction with challenges of using WMATA’s API, a strong subplot of the night was a gentle skewering of Metro’s reliability, as most apps focused on predicting or reporting delays and other incidents throughout the system.
Several, such as Rebecca Mills’ WhatsaMATA site, where users can report their experiences on MetroBus, and Gio Gatto’s Waze-for-trains Traze app, were built around the need for a common, reliable outlet and source for rider complaints and feedback. Both covered issues such as late arrivals, bunching or, as Mills demoed, what would happen if the bus were hypothetically full of goats. Traze, like its road-based namesake, allows Metro riders to tag stations with issues and sort them by line.
Other projects and apps sought to make the experience of Metro more predictable by integrating potential causes of and reactions to common service disruptions. Andrew Yue noticed a significant gap between incidents and official system alerts, proven through a study of Metro’s incident reports, rider tweets, and WMATA notifications. As a result, Yue created WMATA Watcher to detect delays within the Metro system. The app studies tweets, their volume, timestamp, and location and uses this to predict if there might be a delay in the Metro. Yue explained that it is not quite accurate enough for a production model, but is already significantly more accurate than random guessing, and shows promise for delay predictions in the future.
Similarly, James Ferrara’s beta Metro Failure Forecast collects weather data and calculates how likely temperature changes are to cause rail problems that would disrupt commutes. With historical weather data and WMATA’s delay reports, the site employs machine learning, in which an algorithm teaches itself to recognize weather red flags, to interpret which incidents happened in which weather.
Several developers have keyed in to the widespread demand to have the status of the entire system at one’s fingertips. Sites and apps like Thomas Solow’s DC Trains project, Ross Filice’s DC Next Bus, Ian Dixon’s DC Metro and Bus, and James Pizzurro and Jennifer Hill’s MetroHero offer clear looks at delays, arrival predictions, and nearby stops. MetroHero – itself a “control panel”-type depiction of the entire rail system – and DC Trains use a mix of arrival predictions and travel distances to estimate where certain trains are in the system, a piece of information that WMATA will eventually improve upon with its API changes. Different visual depictions, too, of the system help users understand where they can currently find trains and where service issues may be.
On the hardware-display side, Techies debuted its first wearable project with Alex Lindeman’s WMATA With You Pebble watch app, while Tim Burke explained the multi-modal transit dashboard he created for residents of his condo building.
Data reliability is one issue that developers run into, according to Pizzurro and Ian Dixon, creator of the widely-used DC Metro and Bus app. As their dashboards attempt to track the trains moving through the system, many trains often disappear during single tracking or at the ends of their lines. In addition, Dixon raised issues with past data-request limits from WMATA and data showing phantom buses roving through the system.
As a parting comment regarding improving data collaboration, Vavasour emphasized to the crowd that, while there are still problems with the data that WMATA provides, the best way to improve it is to build an app. Putting the data through the trials of real-life use helps WMATA identify problems and find ways to address them. If you have a transit app idea, WMATA wants you to run with it.
Photo: Top, WMATA’s Michael Eichner presents on the agency’s upcoming collaboration with developers (M.V. Jantzen, Flickr, Creative Commons).
See more of what attendees had to say on our Storify page.