6:30 pm – 7:00 Networking and snacks
Both are free and open to the public
Location: GWU Duques Hall Room 451
2201 G Street, NW · Washington, DC
Adjacent to Funger Hall and near the Marvin Center. Parking is available at the Marvin Center. Contact: Dr. Steven Wilcox, 202-497-5774
Predictive Analytics for Policing
John S. Hollywood, Ph.D.
The use of predictive analytics (or, alternatively machine learning and/or artificial intelligence) in policing has been alternately hyped and feared as Minority Report come to life – a means of both preventing crime and threatening the public's privacy and civil rights. In this presentation, we will talk about what predictive policing is – the use of statistical models for identifying places, times, and people at greater risk of criminal involvement – and what it is not, starting with it not being a crystal ball. We will prevent major examples of predictive policing applications and tools, including both the Chicago Police Department's model to predict those at highest risk of being killed and the commercial PredPol tool for predicting small crime hot spots. We will also talk about emerging trends in predictive policing, notably (1) exploiting deep learning and imagery and video analytics to increase the efficacy of surveillance cameras, (2) developing better privacy and civil rights protections, and (3) using predictive policing tools to improve both individual agencies' performance and the field of policing more broadly. We will conclude with a discussion of where these tools might be most useful in both improving public safety and protecting privacy and civil rights.
Dr. John S. Hollywood is a Senior Operations Researcher at the RAND Corporation, where he applies OR to security policy, including criminal justice, homeland security, and defense. He is one of the country's foremost experts on the use of predictive analytics in policing. He supervised the writing of a major reference, Predictive Policing: The Role of Crime Forecasting in Law Enforcement Operations, and co-led major evaluations of predictive policing experiments in Chicago, IL, and Shreveport, LA. He has given interviews for publications including Science and the New York Times, served on multiple expert panels, and given multiple presentations at law enforcement and operations research conferences on the topic.
Outside of predictive policing, he developed methods for making sense of inconsistent reports about potential threats using nonlinear programming and co-led studies of how US terror plots have been foiled; the latter discovered that "over eighty percent of US terror plots are foiled domestically," a fact widely circulated in the counterterrorism field. He was the founding President of the restart of the Public Programs and Processes Section (now combined with two other initiatives to be the Section on Public Sector OR), and currently serves as a section representative to the INFORMS Subdivision Council. He earned a Ph.D. in OR from the Massachusetts Institute of Technology.