Big Data Hits the Road
Big Data Hits the Road
March 2014 | By Craig Harris
The insurance industry has used predictive analytics for underwriting, rating and claims for years. Now, a growing number of municipalities are using the same techniques for a different purpose: road safety.
Data is the new game changer in traffic safety and design of municipal roadways, especially at intersections and on major corridors. While much has been written about vehicle design, analyzing digital data is increasingly seen as a critical tool in identifying high-risk locations and making structural engineering changes to improve safety.
Some cities have taken an active role in gathering data, using the latest business intelligence tools to schedule road retrofits according to cost-benefit analysis. And more are looking into providing real-time information for drivers to promote “situational awareness.” Through digital message boards and even wireless devices, cities can flag drivers about certain conditions, such as inclement weather, unexpected traffic back-ups and construction.
The City of Edmonton is a Canadian pioneer in applying data analytics to road safety. It created the Office of Traffic Safety (OTS) in 2006 with the twin goals of reducing collisions and improving the flow of traffic on municipal roadways. The city has seen a drop in collisions per year from just over 28,520 in 2007 to 23,238 in 2012 – a reduction of more than 18%. Moreover, Edmonton has witnessed a decrease of 11,606 in injury and fatal collisions, with an estimated cost savings of $780 million, according to municipal statistics.
“We take an evidence-based approach to traffic safety,” says Gerry Shimko, executive director of Edmonton’s OTS group, which has 27 staff, including a seven-member data analytics team.“ We are using historical data to predict what is going to happen and then introducing counter measures to reduce collisions, injuries and fatalities. To do this, we have to be very diligent in data collection and data analysis.”
In terms of data collection, Shimko notes that the city gathers a range of information from traffic cameras, roadside devices and road sensors that create a “digital footprint.” Then analysts use software to examine various factors, such as rates of accidents, where specifically accidents occurred, time of day, conditions and causation, among others. “At the start, we relied on manual processes to analyze the data, but now we use business intelligence solutions to look at the data, integrate it and use it for decisions,” Shimko says. “We can then put that data on top of a city road map and identify the intersections and corridors that have the higher number of collisions.”
One example that stood out was the prevalence of collisions at intersections with dedicated right-hand turn lanes, with 30% of accidents occurring at these “hot spots.” The concern involved a single geometric standard for all right-hand lanes that likely contributed to higher than average rear-end collision rates, potential reduced visibility for drivers and increased vulnerability for other users (pedestrians, cyclists) at busy intersections.
The solution was to develop alternate guidelines for several different right-hand geometric treatments that road designers could tailor to specific intersections. “We have made infrastructure changes through a targeted program that, for example, alters the angle of the turn,” Shimko observes. “We have seen a 75% reduction in collisions for each one changed.”
Another issue involves traffic cameras that record red light violations at certain intersections. Edmonton has more than 50 “intersection safety cameras,” which monitor both red light running and speeding. However, in some cases the design of the intersection may be a contributing factor to violations, according to Shimko.
“We are trying to reduce those other factors that may have been overlooked,” he says. “For example, we noticed a lot of vehicles were running the red light at a certain intersection, so we looked at it in more detail. We got our traffic engineers involved and we changed the profile of the light at the intersection. There were some visibility issues that contributed to the problem at this intersection.”
Other municipal regions across Canada are embracing data analytics. The Region of Halton in Ontario is taking a “data-driven approach to road safety,” according to Robert Merritt, the region’s traffic operations and safety coordinator. “We get collision data from the police, and we input that into our system. We have data from all four municipalities within the region, and all of that data resides on one system.”
Halton has also implemented red light traffic cameras at two intersections, and is planning to add five more this year. “So far, we have seen an 80% decrease in right angle collisions – the kind of ‘T-bone’ accidents that cause significant injuries, fatalities and damage,” Merritt notes.
The red light traffic cameras are separate from a broader campaign in Halton Region, known as the Comprehensive Road Safety Action Plan. Through this program, the region focuses on education, engineering and enforcement. It uses measurements called “safety performance functions” and conducts “network screening” to identify key locations, such as major intersections or high-volume traffic corridors, according to Merritt.
“This gives us a list of higher-risk locations and we can prioritize based on safety performance,” he says. “Then we can look at engineering changes, which may involve things like pavement conditions (skidding) or advanced turn signals.”
Halton Region has worked with TES Information Technology, a firm that helps cities improve data collection and analysis for road safety. TES president Greg Szrejber says Halton is fortunate to have data residing on one system, which is not always the case in large municipalities.
“To do proper analytics, you need data from three areas – traffic count, collisions and infrastructure,” Szrejber notes. “Several cities have this data in different systems, so it is hard to put together. We integrate all three into one corporate database, and this can then be analyzed through software tools to make improvements to road safety.”
Szrejber explains that once the data is together in one source, “you can do all sorts of different analysis – by historical records, by types of locations, such as intersections with stop signs versus signalized traffic lights, by types of collisions, such as right angle versus rear-end collisions.“
He also notes that more cities are using data analytics not just for infrastructure improvements, but also for core road and highway construction projects. “If it costs $20,000 to put in an advanced left turn lane, but that results in a 20% reduction in collisions, many cities are saying: ‘we should just do it.’ There is that level of input into construction projects,” Szrejber says.
The City of Edmonton has also benefited from an alliance with a technology provider, in this case IBM through its “Smarter Cities Challenge.” Launched by IBM’s Corporate Citizenship division in 2010, this is a competitive grant program in which $50 million in technology and services is given to 100 municipalities worldwide over three years. In Edmonton, IBM worked with its traffic safety office on data analytics and road safety.
In particular, IBM’s focus on road transportation involves “real-time data that’s already being gathered by RFID tags, GPS devices, road sensors and smart phones to paint a very detailed picture of what’s happening across a complex transportation system at any given moment in time,” according to IBM spokesperson Carrie Bendzsa.
“The wave of the future in transportation is to move from today’s reactive model – in which we discover a traffic problem in real time and then scramble to fix or avoid it – to a predictive model, which uses advanced analytics to model what traffic patterns are likely to be in the near future,” Bendzsa notes.
Shimko credits IBM with improved turnaround time in data integration and a focused approach to using – and verifying – one true source of data. From this, the City of Edmonton has become an innovator in driver communication and road safety – what Shimko calls “situational awareness.”
“In Edmonton as in most northern cities, weather is a huge factor, so we are trying to predict how weather will play a role in specific types of accidents,” he says. “On days with precipitation, we see a significant increase in ‘follow too close’ collisions. We want to create that situational awareness and communicate real-time with drivers through digital messaging devices on the roads.”
Another strategy in Edmonton is to share data more openly with other municipal departments and agencies, including police. As part of its “speed management continuum,” the OTS has collected data on licence plates through intersection safety cameras and other digital devices, giving this information on high-risk drivers to police, according to Shimko.
“We see in some of our data a lot of overlap between crime and traffic incidents in high-problem areas or hot spots,” he notes. “For police, they can use this information to allocate limited resources.” Shimko adds that a process known as Data-Driven Approaches to Crime and Safety (DDACTS), which integrates location-based crime and traffic collision data, is becoming more prevalent in U.S. cities and will likely emerge in Canada in the near future.
Police are not the only group interested in data analytics and road safety. Szrejber notes that organizations, such as the National Highway Traffic Safety Administration in the U.S., are publishing guidelines on data and road safety improvements. These manuals are setting standards for road design – something that lawyers have noticed.
“Lawyers are increasingly researching safety manuals and using them in collision-related lawsuits,” Szrejber says. “They are asking cities if they follow the guidelines and holding them accountable. This is a significant liability issue.” Another potentially interested party may be the insurance industry, which could use road safety and collision statistics as underwriting and rating factors for auto insurance in municipalities and regions.
Is it possible to align insurance premiums, at least to some extent, with the degree to which cities have adopted the latest safety measures in roadway and intersection design? That is a question with a moving target, as data analytics and traffic safety continue to evolve in municipalities across Canada.
Please note a version of this trends paper has previously been published in an industry publication.
ADVANTAGE Monthly trends papers
This paper is part of an open online library of ADVANTAGE Monthly trends papers, published by the CIP Society for the benefit of its members and of the p&c insurance industry. The trends papers provide a detailed analysis of emerging trends and issues, include context and impact, and commentary from experts in the field.
The CIP Society represents more than 18,000 graduates of the Insurance Institute’s Fellowship (FCIP) and Chartered Insurance Professional (CIP) programs. As the professionals’ division of the Insurance Institute of Canada, the Society’s mission is to advance the education, experience, ethics and excellence of our members. The Society provides a number of programs that promote the CIP and FCIP designations, continuous professional development, professional ethics, mentoring, national leaderships awards, and research on the issues impacting the p&c insurance industry in Canada.