Predictive Enforcement of Pollution and Hazardous Waste Violations

Fellows: Benjamin Brew, Jane Zanzig, Asif Zubair
Data Science Mentor(s): Eric Rozier
Project Manager: Paul van der Boor
Project Partner: U.S. Environmental Protection Agency

Two federal laws regulate the the treatment, storage, and disposal of hazardous waste in the United States: the Resource Conservation and Recovery Act (RCRA) and the Risk Management Plan (RMP). To enforce these regulations, the Environmental Protection Agency (EPA) or state agencies regularly conduct inspections of facilities that handle hazardous materials. Going forward, the EPA wants to conduct more targeted investigations, using historical inspection data to predict the risk of severe violations.

Using EPA data sources on reporting, monitoring, inspection, and enforcement from the RMP and RCRA databases, the  2015 DSSG team developed and evaluated predictive models to identify likely violators. These scoring models are weighted by various criteria of importance to the EPA; including the potential magnitude of the violations, the environmental and public health impact of violations, and the likely outcome of an enforcement action. As a result, the EPA will be able to effectively identify potential violators, better allocate inspection resources, and maximize the impact of each investigation to keep America’s air and water clean.

We expanded on this work in 2016 while partnering with the New York State Department of Environmental Conservation. You can read more about our work on environmental conservation here.