SafeRoad: A Data-driven Approach to Safer City Streets
According to the Centers for Disease Control 2014 Final Report, traffic fatalities are the leading cause of death for persons under age 35 and the third leading cause of unintentional death amongst those under age 55. Regardless of prevalence, we believe that all traffic accidents are avoidable; which means no traffic fatality is acceptable.
There are several programs across major US cities aimed at using data-based decisions to design safer streets and educate the public to reduce traffic injuries and fatalities. With our product, however, we look to fill a tactical gap, providing near-real-time prediction of traffic collisions that will show public leaders and traffic safety specialists when and where serious collisions are more likely to occur. Our tool, SafeRoad, is a big data system for large-scale collision data mining and prediction. Our tool will identify what road and other conditions are important causal factors for fatal and serious injury collisions to develop long-term, city-specific strategies to meet the goal of zero traffic fatalities.
SafeRoad is a fully-automated, end-to-end collision analysis and prediction tool built entirely on open-source, big data technology and freely available data sets. In order to model the complexity of automobile collisions in a large metropolitan city street system, several datasets are collected and used as input to the SafeRoad model.
Our proof-of-concept system utilizes historical automobile collision data retrieved from the New York City Vision Zero project that contain geo-coded records for all automobile collisions reported to the New York Police Department dating back to July, 2012. Additional data are used to enhance the accuracy of the model, such as weather retrieved from the National Oceanic and Atmospheric Administration (NOAA), and other city and state-specific data sets made available through the New York State and New York City open data initiatives.