20140905-142852-47

Introduction
This description will provide background and details on a methodology developed at the U.S. Naval Research Laboratory
(NRL). This methodology has shown applicability to modeling criminal events and will be demonstrated to
the Detroit Police Department (DPD) using their crime events to evaluate the technology. Finally, the technology
will be tested in an operational use case. DPD will be able to anticipate events and deploy resources to areas identified for future crime.

Background
The NRL has developed an advanced geospatial methodology to characterize a set of events by combining geospatial
and statistical techniques to quantify “place” as a measure of spatial preference for target selection. With a model for
preferred place, areas that match the original signature can be highlighted. Using this methodology, in conjunction
with other analytical tools and information, one can more fully understand the events being characterized and take
preemptive action.
The methodology was initially designed to quantify the spatial characteristics of Improvised Explosive Devices in
Iraq and was subsequently used to explore spatial-patterns of Person-Borne Improvised Explosive Devices (PBIEDs)
in Israel. To further validate the methodology using richer datasets, a series of experiments were conducted in
collaboration with the Cambridge, MA Police Department, using subsets of their crime data. In the course of the
experiments, the methodology was applied, with some success, to commercial burglary and robbery series and street
robberies.
A range of literature from across disciplines were consulted when formalizing the model. The theories draw
from epidemiology, criminology, and ecology. An example from criminology is situational crime prevention, which
postulates an event occurs when there is an intersection among: an offender, the victim, and a place. By applying
spatial-profiling techniques, “place” can be better understood, which can be applied to address ongoing and emerging
crime problems.
Planned Usage in Detroit
Proof of Concept
NRL will first work with DPD to demonstrate utility of the methodology on Detroit’s data. For this, it will be
necessary to have a number (approximately 5-10) of distinct groups of events (i.e., series crimes), ranging in type
of crime, to analyze.

Brief
The results from the proof of concept will be presented, in-person, in Detroit. Feedback will be elicited from DPD
and other involved parties. Additionally, a plan for an operational case study will be developed.

Refinement
Using feedback from the brief of the proof of concept, NRL will refine the process and tailor the results to better meet
DPD’s needs. Depending on the readiness of the technology, additional models can be generated from historical or
on-going events.

Operational Case Study
Ultimately, the capability will be tested for validity in an operational setting. This will entail the use of on-going
events as input to the model. Results will be run and results shared with DPD in a timely manner.

Moving Forward
If the model proves to be operationally valid, resource permitting, a framework will be developed to fit the methodology in alongside DPD’s extant analytical processes, including capabilities and other sources of information. Once these capabilities are tested they can be shared with other state and local jurisdictions.