Transportation Data Science at Microsoft

Published July 13, 2017

Share this page

West Hub Steering Committee Member Professor Kristin Tufte moderates a panel at the launch of the National Transportation Data Challenge in Seattle, May 2017.

By Vani Mandava (opens in new tab), Director, Data Science Outreach, Microsoft Research

The National Science Foundation (NSF)-supported Big Data Innovation Hubs launched a National Transportation Data Challenge (opens in new tab) with a kickoff event in Seattle in May 2017. Microsoft Outreach, through its partnership (opens in new tab) with the Big Data Hubs organized an Azure workshop and participated in a panel discussion (opens in new tab) on ‘How Cloud Computing Can Enable Transportation Data Science.’ The kickoff was the first in a series of events that are being organized across the US to launch this challenge. It is an activity that spans all four hubs, and is expected to reach all 50 states. Several teams across Microsoft contributed ideas on recent or ongoing work on transportation data science. Below is a summary of the all the contributions that were part of the event.

Microsoft’s engagement with the Challenge builds upon a foundation of prior work in public safety and metro data science. The Challenge launch event highlighted a collaboration between Microsoft’s Civic Technology Engagement (CTE) group within the Corporate, External and Legal Affairs (CELA) team and DataKind (opens in new tab), Vision Zero (opens in new tab), the New York City Department of Transportation, Seattle Department of Transportation, and the City of New Orleans’ Office of Performance and Accountability. The project enabled an ecosystem that helped cities (opens in new tab)assign limited resources to prioritized traffic safety issues. Adam Hecktman and Kevin Wei from the CELA CTE team also built a cool interactive Power BI dashboard that demonstrates and visualizes 300M+ bike rides in the city of Chicago (opens in new tab).
Microsoft Research’s Video Analytics Towards Vision Zero (opens in new tab) was represented on the panel by Franz Loewenherz, City of Bellevue, and was mentioned by both Daniel Morgan (Chief Data Officer, USDOT), and former governor, Chris Gregoire. On June 1^st, Bellevue officially launched the Video Analytics Towards Vision Zero crowdsourcing initiative. In a collaboration with organizations across North America, Bellevue, the University of Washington and Microsoft are asking for the public’s help analyzing traffic camera footage to teach computers how to identify and track people using wheelchairs, bikes, and other modes of transportation as they navigate intersections. The more people who go online, the better we can “teach” computers to scan traffic videos and recognize near-collision events (see City of Bellevue Media Release (opens in new tab)). Microsoft Research scientists leading this effort are Victor Bahl (opens in new tab) and Ganesh Ananthnarayanan (opens in new tab).
Wee Hyong Tok, Principal Data Science Manager in the Cloud AI Platform group built an Azure Machine Learning based predictive model for incident severity reporting based on the National Highway Traffic Safety Administration (NHTSA) Fatality Analysis Reporting System (FARS) data. The model has an accuracy of 68% and can be used to provide a baseline model (opens in new tab) for participants. Additionally, Patrick Baumgartner and the PowerBI team built a compelling interactive PowerBI visualization (opens in new tab) based on the dataset not only demonstrate analyses, various correlations but also dive deeper into visualizing point of impact and seating position.
Transportation Data Science efforts extend beyond the United States. Andrew Bradley, Principal Solution Specialist on the Microsoft UK Enterprise and Partner Group (EPG), shared how the UK team is engaged with the Department of Transport, UK, and are actively encouraging innovation in the region by supporting events, hackathons and challenges.

Microsoft’s Connected Vehicle Platform (opens in new tab) recognizes the digital transformation that is reshaping the automotive industry (100% of new cars by 2030 are projected to be connected) and is investing in building extensible, global, and scalable automotive solutions in partnership with organizations such as Nissan, Volvo, and BMW.

We look forward to engaging with the transportation data science community as the National Transportation Data Challenge takes shape over the coming months.

Learn more