Hot spots in Louisville | “Automobile Collision Prediction in Louisville, KY”

A new data project has discovered the most dangerous intersections in Louisville, and its implementation could point the way toward a safer streetscape — for motorists and pedestrians alike.

The project, “Automobile Collision Prediction In Louisville, Ky,” was conducted by students in the University of Pennsylvania’s Urban Spatial Analytics Program through a data-sharing partnership with Louisville Metro Government’s Office of Performance Improvement and Innovation. Louisville was one of five cities selected to participate in the study.

The analysis comes at a time of increasing pedestrian deaths in the city and a rapidly changing urban core, the latter due to the completed Ohio River Bridges Project, multiple large-scale construction projects, and the promise of a new bus-rapid transit system.

Ken Steif, a 20-year data analytics veteran and the professor who heads the UPenn Urban Spatial Design Master’s program, said that high instances of traffic collisions aren’t unique to heavily car-dependent cities like Louisville.

“This is very typical in America,” Steif said. “Roads that are very fast, with many lanes. And that sort of atmosphere is not conducive to pedestrians; it puts them at great risk. It also factors in terms of the sheer number of crashes that occur, not only car to car, but car to person.”

“But the real goal,” he continued, “is to use data to design smarter, safer spaces.”

Via their collaboration with Louisville’s open data program, Steif’s students fed city traffic data into a sophisticated prediction model, incorporating a wide variety of variables — everything from traffic speed, land features, traffic lights, proximity to downtown, potholes and more.

As a result, they were able to identify collision “hot spots” in the city.

Those hot spots identified by the study and using Kentucky State Police traffic collision data are Bardstown Road at Grinstead Avenue, and in Downtown Louisville, as shown below:

“Automobile Collision Prediction in Louisville, KY” by Guy Duer, Sha Ni & Deon Provost, UPenn

Michael Schnuerle

Louisville Metro Chief Data Officer Michael Schnuerle said that he and Metro Innovation Project Manager Ed Blayney worked with their respective staffs to provide as much data as possible to Steif’s team to create a data-based a road safety plan for the city.

“This work with UPenn, the way I look at it, in the end it’s a free data analysis that we can use,” Schnuerle said. “And it only came about because of the Office of Civic Innovation, and that was only possible because the mayor made it possible in the first place.”

John Callihan, Louisville Metro’s Director of Transportation, said that the problem of traffic collisions “is a community livability and public health issue.”

Callihan, whose background is in engineering, said that his department has been busy trying to spend $20 million in road paving over the past year. He said that similar, low-tech solutions can help implement the study’s findings into concrete solutions.

“One thing we might be using this information to consider is, do we need to take a closer look at the speed limits at a particular segment of roadway?” Callihan said. “It’s what we call traffic calming: what can we do to influence drivers’ behavior. That can be anything from changing a traffic signal controls, or changing lane widths, how lanes are assigned, sidewalk spacing, and more.”

As far as the more sophisticated — and, therefore, more expensive — solutions, Callihan cites initiatives already in the planning stages, like laying fiber optic cables into major traffic corridors to help create “smart box” traffic intersections that can dynamically adjust to flows of traffic throughout the day.

He said that the newfound crosstalk between civil engineers and city data officers is making it possible to tackle problems like traffic safety in new ways.

“We’re starting to understand that it’s not just about moving a car from point A to B,” Callihan said, “but how mobility fits within the greater context of our community.”

Schnuerle said he’s considering ways for Louisville to accommodate driverless cars, but that the reality of AI-controlled cars rationally avoiding collisions on the city’s streets is still a way’s off. In the meantime, the city’s partnership with WAZE, a real-time traffic information service, is providing a stream of traffic data to any motorist who downloads the app onto their smartphone.

But with WAZE and the ubiquity of Google Maps funneling every driver who uses them onto the same, “best” route, such app-eccentric solutions can cause the very problem they’re trying to solve.

“WAZE and Google and other routing companies have come under fire for specific situations,” Schnuerle said. “For example, say in some city there’s an accident that reroutes everyone through a quiet suburb and it can’t handle it. It creates more congestion.”

But Schnuerle said that by providing the most up-to-date information on construction zones or accidents to WAZE, the software can more quickly react to changing road conditions.

“[These apps] all work in real-time to adjust,” he said. “So if there’s an accident on the expressway, 30 percent of drivers might get off, but that causes a backup. Yet once it tips back over, the app can put you back on that expressway once it’s faster again.”

One of the most obvious ways for cities to reduce congestion is to provide transportation options such as light-rail that get cars and pedestrians off the street. But Callihan said there were no plans to implement such a system.

“Using data to squeeze the most out of our infrastructure is true,” Callihan said. He acknowledged that the argument against cities like Louisville being unable to support rail-based transit systems are financial in nature, he said the first major short-term solution is implementing the state’s first bus-rapid transit line along Dixie Highway.

The $50 million, TIGER-grant-funded “New Dixie Highway Project” will be operated by the city’s public transportation provider, the Transit Authority of River City, beginning in 2019 to help ease congestion on the major road artery.

The BRT-line is one of a handful of road improvement projects actually under construction as part of the city’s “Move Louisville” 20-year transportation plan. As of a March 2018 progress report, most of those projects are still in planning stages, and a few are slated to begin work later this year.

The city’s new rapid transit line will have quite the burden to bear by the time it comes online. TARC has reduced route services over the past decade, and is currently experiencing a driver shortage. And that’s despite evidence that providing robust public transportation options can reduce traffic crashes by as much as 90 percent.

One of the UPenn study’s more notable findings is that traffic collisions in Louisville rose by over 37 percent from 2013-2016.

Pedestrian-related deaths increased to 30 in 2017 from 24 in 2016, according to Kentucky State Police data analyzed by Insider Louisville. And that increase occurred despite a slight decline in overall pedestrian-related injuries during the same period, with 350 in 2017 compared with 376 in 2016, according to the KSP data.

In 2017 alone, the KSP data reveal that 105 fatal auto accidents occurred in Jefferson County, with nearly a third of those pedestrian-related. And since the beginning of 2018, there have been 12,029 collisions — down from 13,088 from the previous period last year — with a total of 21 fatalities. Eight of those deaths included pedestrians.

“In, Philadelphia, my home, the emphasis [in city planning] tends to be on pedestrians,” Steif said. “But cities in the northeast, in Philly, tend to be older, and are more transit-dense cities. Relatively newer cities like Louisville are more dependent on cars. Information and analysis like this can be used to design safer and more effective space where cars are going to share space with pedestrians.”

But no model, Steif acknowledged, is perfect.

“Nobody has a crystal ball,” he said. “And anybody who uses data and says they can predict something 10 years down the line, they’re not being forthright. But you can use data to predict the near-term.”

The biggest revolution in data, he concluded, doesn’t lie in specific modeling techniques or ultrasophisticated technologies, but in something decidedly more old school: sharing information.

“The biggest innovation is the open data movement, and pretty much every large or medium city is putting non-private data out there, fostering collaboration,” Steif said. “In my mind, that is a real revolution, a real high point in tech innovation in the last two or three decades. And it starts with governments and mayors and city councils, etc., embracing evidence-based public policy.”

Schnuerle said that having this data is just the beginning of a longer process to reshape the city into a safer and more livable place.

“The data side of it, this is a nice start,” he said. “But after that there’s still a lot more to do.”