About Welcome to the webpage of the NSF Operations Engineering (OE) Program Workshop on Decision Analytics for Dynamic Policing. This workshop is funded by NSF (NSF CMMI-1917624) to explore research directions in decision analytics and operations research (DA/OR) for enabling improved dynamic policing strategies from a variety of aspects. NSF program directors will be welcome to attend at any time during the workshop. In these pages, you will find general information about the workshop (below) and information on workshop agenda, workshop organizers, keynote speakers, and recommended reading. All workshop attendees should review The University of Texas at Arlington’s code-of-conduct policy. The workshop code of conduct with follow this policy.
All workshop attendees should review The University of Texas at Arlington’s code-of-conduct policy. The workshop code of conduct with follow this policy. https://www.uta.edu/policy/hop/5-513
Date and Location
Date and Location Date: May 9-10, 2019
Location: Renaissance Arlington Capital View Hotel, Arlington, VA
Project Summary and Report
The traditional approach in law enforcement studies utilizes randomized experimental design. However, this approach does not take advantage of today’s data collection capabilities. Within the last decade, the concept of predictive policing was proposed by a RAND report, funded by the National Institute of Justice (NIJ), to categorize and encourage academic research developing machine learning and statistical modeling tools to accurately predict crime. Unfortunately, existing predictive policing methods are flawed in four critical aspects: (1) They omit the human perspective, including human behavior, political biases, and community perception. (2) They assume that the data include the necessary information for both prediction and action. (3) They assume that historical data on the policing system will be representative for the future. (4) They assume unlimited police resources.
A key element in the RAND report is the concept of prediction-led policing. For this concept, it is recognized that prediction alone is inadequate without connection to a policing strategy that identifies appropriate actions. This proposed workshop on decision analytics for dynamic policing organized local teams of police departments (PDs) and academics with expertise in DA/OR and criminal justice (CJ) to build a cross-disciplinary understanding of policing challenges that may be addressable by DA/OR. Discussions took place within and across these populations, so as to examine all three perspectives.
DA/OR as a discipline is not well known within the law enforcement context. Towards filling this gap, the workshop identified and recommended new directions for research in DA/OR to develop methods for policing that are cost-effective, consider social influences, identify the appropriate data, address the dynamic nature of policing, and utilize existing knowledge from PD/CJ expertise. Overall, the potential impact of DA/OR research in law enforcement is extremely high, but progress will require care, collaboration, persistence, and an open mind.
The objectives of the proposed workshop are:
- Build an understanding of the dynamics of policing systems.
- Build an understanding of the controllable decisions and uncertain elements in dynamic policing systems, for example, intervention teams, criminal responses to police actions, community outreach, interactions between PDs and disciplines related to policing (such as forensic science divisions), etc.
- Create processes that facilitate research partnerships between DA/OR, CJ, and PDs to encourage the development of DA/OR methods for dynamic policing systems.
- Identify critical research themes in dynamic policing that may be studied using DA/OR.
- Identify methodological advancements needed in DA/OR to address the critical research themes in dynamic policing.