Predictive Policing Symposium: Discussion on the Predictive Policing Demonstration Projects and Evaluation
Boston Police Department
In 2009, seven police agencies received planning grants through NIJ's competitive solicitation process. At the First Predictive Policing Symposium, representatives from six sites introduced plans for their projects. The analytic support and evaluation grantee, the RAND Corporation, also discussed its role. The Los Angeles Police Department is the seventh site. For information on its plans, see the session on The LAPD Experiment.
Hemali Gunaratne explained that the Boston Police Department plans to develop, test, implement and evaluate a predictive policing model for property crime, which comprises 77 percent of the city's crime. The goal is to develop a model to be used in management planning activities, officer deployment and COMPSTAT.
Cmdr. Jonathan Lewin said that a primary goal for the Chicago Police Department is to produce technology and products that officers on the street can use. The focus of the grant is to build on several efforts already in progress, including work with SPSS/Spadac, Oracle and Carnegie Melon University.
Philip McGuire noted that since 2001, the New York Police Department (NYPD) has replaced its record management system, constructed a data warehouse, provided more online access and significantly upgraded various tracking systems. These changes will facilitate predictive policing implementation. NYPD will use its grant to review all competing analytic options and decide on a set to employ under the predictive policing umbrella. The department will then analyze the effect on operations.
Speaking for the Maryland State Police Department, Philip Canter explained that the goal of the program is to use analysis tools, technology infrastructure and governance to ensure all analysis is supported. Dr. William Pottenger explained that his focus for the project was on analysis tools and the technology infrastructure. Currently, he is looking at Plug 'n Play open-source Web 2.0 framework.
Representing the smallest department of the seven in the demonstration projects, Susan Reno explained that the Shreveport Police Department is developing an in-house project using out-of-the-box software. A goal for this project is to design a predictive model for tactical crime and police districts based on previous research. The project will test the hypothesis that a group of suspicious and disturbance calls is a leading indicator for "tactical crime" (e.g., shootings, robbery, burglary and auto thefts).
Brenda Eich described the strategic crime briefings conducted by the District of Columbia Metropolitan Police Department. With the NIJ grant, she said the plan is to pattern crime by time and space to be more predictive. The first phase of the project is to compile information by using multiple statistical models during development. The goal is real-time forecasting.
Dr. Greg Ridgeway said that in the planning phase over the next year, RAND will provide technical assistance to the seven sites by providing analytical support. In future years, RAND will ensure data collection abilities are available. Other considerations for RAND include:
- Establishing a scientific steering committee to develop logic models for each site.
- Reviewing and refining implementation, data collection and evaluation plans.
Update: In 2011, two of the original seven sites (Chicago and Shreveport) won competitively awarded grants to continue into Phase 2 of their demonstration and evaluation of predictive policing strategies. The Chicago Police Department was awarded $1.5 million. The Shreveport Police Department was awarded $516,474. Both Departments are working with researchers to evaluate the strategies they developed in Phase 1. Chicago's efforts are focusing on gang conflicts. Shreveport is focusing on tactical crime prevention.