Predictive analytics at work: predicting traffic jams before they occur

Starting next year, drivers on the New Jersey Turnpike will receive up to a 10-minute advance warning before traffic backs up.
Written by Joe McKendrick, Contributing Writer on

They call the New Jersey Turnpike America's "Muscle Road," and with good reason. It directly connects the largest metro area, New York, with the nation's fourth largest, Philadelphia, and also serves as the gateway south to Baltimore and Washington, DC.  As it nears the New York region, it expands to 12 lanes, cutting through a colossal industrial landscape of refineries, factories, office buildings, airports, and port facilities. Close to 250 million vehicles travel the road every year.

Now, Bloomberg's Dunstan McNichol reports, the agency that runs the turnpike has awarded a $652,000 contract to En Pointe Technologies Inc. to install technology that will detect impending traffic jams and alert drivers to divert away from the impending bottleneck -- 10 minutes before the event happens.

The system has already been tested on the Garden State Parkway, another traffic magnet (especially on summer weekends), and achieved 93 percent accuracy in predicting traffic jams, Brian Gorman, director of technology for the New Jersey Turnpike Authority, is quoted as saying. He believes an accuracy rate of 90 percent can be achieved on the turnpike. The technology was developed by IBM, and has been already been employed on highways in Singapore and Stockholm (covered here at SmartPlanet).

The traffic prediction software, which will deliver alerts via electronic signs along the turnpike and Garden State Expressway, will be put into production early next year.

NJ.com and Newark Star-Ledger report that the Turnpike Authority says that traffic forecasts would possibly resemble color-coded forecasts, with green roadways meaning a delay of five minutes, green with yellow a delay of 10 minutes, and yellow with a lot of red meaning 15 minutes.

(Photos: New Jersey Turnpike Authority.)

This post was originally published on Smartplanet.com

Editorial standards