The Transportation Data Challenge technology sponsorships provide executable templates and examples of transportation data science applications based on open data. Participants can submit their work for the showcase, and if accepted it will be hosted by the technology sponsors and leverage any number of available technologies. Resources provided thus far from companies including Microsoft, Google, Amazon, Waze, Satori and DataScience.com.
Below is a selection of data science demonstrations featured as part of the Transportation Data Challenge.
simulating mixed human and automated traffic
How will human and automated automobiles interact with each other? What's the impact on traffic and infrastructure planning? This team from U. Illinois leverages python scripts and traffic simulation software to map out a variety of scenarios when bots and humans share the road.
Analyzing cycle Share data with Python and pandas
Pronto CycleShare in Seattle provided an open dataset and U Washington Professor and data science wizard Jake Vanderplaas demonstrates how to estimate trip distances, rider speed, weather impacts, and more from this open dataset.
CAUSAL IMACT ANALYSIS
LA-based data scientist Ben Van Dyke uses the causal inference technique developed by Google scientists and applies it to open traffic data from the induction loops embedded in the roads of LA. This executable demonstration provides examples of how to create a counterfactual experiment and leverage a Bayesian structural time-series model to evaluate the impact of a real infrastructure intervention. The analysis was also featured in an editorial in the LA Times.
analyzing distracted driving with tensor flow
State Farm Insurance sponsored a series of challenges using distracted driving experimental data and asking volunteer data scientists to write scripts to analyze video and create 'distracted!' alerts. This example using Python and TensorFlow leverages neural networks and supervised learning to model an accurate alert system based on video feedback.
Standardizing Sidewalks for a Global Pedestrian Data Network
Sidewalks provide a primary mode of travel that supports nearly all other travel options, leisure, recreation and community activities. This project out of the University of Washington seeks to make pedestrian ways, particularly sidewalks, first class members of an open data transportation network.
additional projects to be featured
- Modeling Bike Sharing in R
- Traffic Resilience with Autonomous Vehicles
- Rebalancing CitiBike
- Predicting Out-of-Service Buses
- Accessing and Calculating the Los Angeles Vision Zero High Injury Network
- Mobilizing teens to use open data to learn data science in Los Angeles
- Leveraging Waze Community Partnership JSON Feeds for Traffic Integration
- Aggregating Streaming Bicycle Data Sets with Real Time Analytics using Satori