The Department of Transportation's (DOT) Safety Data Initiative seeks to integrate data sources with each other and with new 'big data' sources that are becoming available to enhance our understanding of crash risk and our ability to mitigate it. The initiative seeks to build DOT's capacity to translate the successes of predictive data analytics tools used by private industry and universities to identify systemic factors contributing to serious crashes.
Evolve from retrospective to predictive analysis that is transformed into compelling visualizations of insights to better target safety risk. Develop an integrated data eco-system for rapid, rigorous, and innovative safety data analysis and insights using new datasets and new analytic tools across surface transportation modes, accessible to decision-makers to support policy decisions. Be a leader in Safety Risk Management by working with transportation stakeholders and leveraging innovative data sources and practices.
We will strategically prioritize and address transportation safety risks through data-informed decision making, with a focus on:
- Data Visualization: Make data analysis and insights accessible to policy-makers through clear, compelling data visualizations.
- Data Integration: Integrate existing DOT databases and new private sector data sources to answer safety questions.
- Predictive Insights: Use advanced analytic techniques to identify risk patterns and develop insights that anticipate and mitigate safety risk to reduce injuries and fatalities.
In order to achieve this vision, we are pursuing three strategies:
- Build DOT’s capacity to perform data analysis for policy and decision making based on risk and predictive insights;
- Establish data integration inside DOT and through collaboration with other agencies and entities to create data connections and integration; and
- Promote the innovative use of safety data and visualization among traditional and non-traditional stakeholders to turn data into useful information for continuous safety improvement.
DOT launched a safety challenge asking participants to come up with innovative ways to visualize data that will reveal insights into serious crashes on our roads and rail systems while improving our understanding of transportation safety. Five Stage I semi-finalists are advancing to Stage II, developing their ideations for an analytical visualization tool into a proof of concept. For more information, visit the Solving for Safety Submissions page or read the press release announcing the five semi-finalists.
DOT’s Volpe National Transportation Systems Center ("the Volpe Center") is leading a pilot project exploring the opportunity to estimate police-reported traffic crashes in near-real time by combining crowdsourced crash data from Waze with crash data provided by the State of Maryland via the National Highway Traffic Safety Administration’s (NHTSA) Electronic Data Transfer pilot. The Volpe Center employed machine learning techniques with these datasets to train statistical models to predict crashes. In this pilot, DOT learned these models supported with Waze data produce reasonably good estimates of police-reported crashes. This pilot has laid the foundation needed for a future nationwide scale-up of a crash count tool.
The rural speed pilot is an ongoing research effort to understand the contribution of prevailing speed, speed limit, and average travel speed to the prevalence and severity of crashes on rural highways. The pilot further seeks to understand the relationship roadway design and traffic volumes have with speed and crash outcomes. The Federal Highway Administration (FHWA) is using the National Performance Management Research Data Set (NPMRDS) – anonymized data from GPS-enabled devices – to learn about speed’s role in crashes. The NPMRDS provides prevailing speeds at 5-minute intervals across the entire National Highway System. When combined with traditional datasets, the NPMRDS data can provide a closer look at speed’s role in crashes.
The pedestrian fatalities pilot sought to understand the relationship pedestrian fatalities may have with transportation system and built environment characteristics. Two key takeaways were discovered through analysis of data from FHWA, NHTSA, the Environmental Protection Agency (EPA), and the U.S. Census Bureau. In urban areas, traffic on non-access controlled arterials was found to significantly increase pedestrian fatality risk. Traffic on other urban roadways and all roadway types in rural areas also contributed to pedestrian fatality risk, but with weaker effects. Additionally, employment density in the retail sector was strongly associated with increased pedestrian fatality risk in both urban and rural areas. Lessons learned from this pilot may be used to understand place-specific risks. The work from this project was published in the December 2018 edition of Accident Analysis & Prevention.
NHTSA is experimenting with the presentation of its Fatality Analysis Reporting System (FARS) data – a nationwide census of fatal injuries suffered in motor vehicle crashes – to supplement existing data summaries on specific topical areas. NHTSA is in the process of beta testing an interactive visualization of the 2016 Traffic Fact Sheet focused on speeding using visualization software. By creating more interactive information, the hope is to present the data in a new way that may be helpful to policy-makers and the general public.