Description:
The Seattle Department of Transportation is working internally and with Microsoft, the nonprofit DataKind, and data scientists from University of Washington, Microsoft Research, HERE Maps, and others to use data to predict crash probabilities, with the goal of reducing bicycle and pedestrian fatalities and serious injuries to zero.
Challenges and Solutions:
Seattle’s Vision Zero Plan lays out an aggressive goal to reach zero traffic-related deaths and serious injuries by 2030. More than 30 crashes occur every day in Seattle and transportation-officials generally evaluate crash locations for safety improvements after the fact. Through this research, Seattle intends to create predictive models that will allow the City to take a systemic approach and proactively to improve safety.
Major Requirements:
1. Collect and clean datasets.
2. Identify factors that commonly contribute to collisions.
3. Apply findings citywide to create predictive models.
Performance Targets/ Key Performance Indicators (KPIs):
• Eliminate transportation-related serious injuries and deaths by 2030
• 10 percent reduction in fatalities and serious injuries annually
Measurement Methods: Describe the methods to measure the performance/KPI impact to assess the benefits to the residents/citizens.
• The Seattle Department of Transportation (SDOT) will continuously maintain collision records and post progress toward our goal on a public-facing performance dashboard.
Standards/Interoperability:
Cities often have similar infrastructure challenges. By understanding which factors are the greatest predictors of crashes, any city can improve its urban design for safety.
Replicability, Scalability, and Sustainability:
The collision factors identified through data-analysis could easily be shared with other government entities for use at the local, state and federal levels. In addition, SDOT will develop a model to refresh output annually.
Impacts:
AAA and the Centers for Disease Control and Prevention estimate that each serious collision costs society approximately $6 million in the deployment of emergency resources, traffic congestion, insurance claims and lost wages. For Washington State, that translates into about $700 million annually. Reducing crashes will save lives, reduce congestion, lower the monetary and reduce burdens on first responders.
Demonstration/Deployment Phases:
Phase I Pilot/Demonstration June 2016:
DataDive with DataKind, Microsoft, and University of Washington: Presenting initial results
Phase II Deployment June 2017:
Using the findings from our research partnership, implement predictive collision modelling and begin applying engineering countermeasures based on outputs.