Spring 2011

The Newsletter of the Minnesota GIS/LIS Consortium

Table of Contents

MN GIS/LIS Consortium

From the Chair
Conference Planning

Mn/DOT Road Closures Site
Economic Data
Gypsy Moth Response

Proximity Finder

Red River LiDAR

Minneapolis Predictive Crime Mapping
Anoka Co. Recreation
GIS - Core Govt. Service

Land Cover 2006
USGS Historic Topo Scanning

Higher Education
Smart Growth & Transit

Will Craig, UCGIS Fellow
Robert McMaster, UCGIS Education Award
Marv Bauer, Pecora Award


Advanced Analysis using GIS at the Minneapolis Police Department
By Ryan Hughes and Sgt. Jeff Egge

“Targeting the Next Crime,” a recent Star Tribune feature by reporter Matt McKinney, highlights the work of the Minneapolis Police Crime Analysis Unit in its mission to identify crime today in order to prevent crime tomorrow. What the article does not fully feature is the pivotal role that Geographic Information Science (GIS) plays in this fledgling pursuit towards more efficient and cost effective deployment of police resources to places where they are needed most.

“Hot Spots” in Minneapolis
In 1988, Minneapolis collaborated with Criminologists Lawrence Sherman and David Weisburd as they conducted a one-year randomized trial of the effects of police in 55 of 100 troubled locations. Their groundbreaking research is the foundation for Minneapolis’ evidence-based policing methodology that focuses on the predictability of places.  Contemporary research has found that police strategies are more effective when they are place-based, proactive and focused (Lum, Koper, Telep).  Since the implementation of CODEFOR, Minneapolis’s version of COMPSTAT, there has been a 34% decrease in crime monitored and measured by crime analysts for over 680 consecutive weeks.  The Crime Analysis Unit’s experience supports the Sherman and Weisburd research that shows places to be six times more predictable than people, and that to complement our intelligence-led efforts, our analysis for patrol should focus on the ‘wheredunit’ more than the ‘whodunit.’

GIS in Predicting Crime
In many ways advanced crime analysis is a logical calculation of symbols, factors, quotients, fractions, and exponents. In an effort to illustrate the consideration of analytic variables, the MPD devised a formula to guide the process of predicting crime. 

The geometric triangle is a known crime analysis model for relating essential variables from each crime. Crime and disorder is an alignment of suspects, victims, locations or a desired commodity. To forecast or predict the product of this association requires a factoring of a wide range of dynamics symbolized by x and y.  X and Y were chosen to epitomize a universal expression that is used to argue function, elements, inputs and outputs in proportion or relative amounts. Most importantly, for students of geospatial technology, x and y are geographic coordinates that help crime analysts map and measure place based interrelationships of crime.

There are multiple spatial analysis tools one can use to predict the next robbery, burglary, auto theft, etc. We have used the CrimeStat - Correlated Walk Analysis (created by Ned Levine and Associates), Probability Grid Method, Geographic Profiling – Rigel Software, ArcMap scripts, Micro Hot Spots and Leveraging of Data.

Figure 1: Part I Crime (Pin Map) occurring between Feb 22, 2011 - Feb 28, 2011.

Traditional pin maps (Fig 1) show individual events over the political boundaries of a city.  Initial attempts at developing hot spots were achieved with snapshot views of using kernel density.  Though kernel density (Fig 2) offers a vivid illustration, it is limited in the capacity of measuring change over time, or more importantly, the displacement of crime to other places.  The challenge was to reduce large concentrations of crime into micro hot spots that are manageable and “bite sized.”  In the spirit of Sherman and Weisburd’s definition of Minneapolis hot spots, the size should be able to be treated by one deployed squad car or beat officer.   For a city which grew into a street configuration that is predominantly north-south and east-west, the use of geographic grids offered the best visual to illustrate problem epicenters of crime. The ideal size of a hot spot grid was established as 1000 by 1000 feet, or (3 x 1 city blocks, .035 square miles) (Fig 3). The grids were created by using Hawth’s Tools “Create Grid” process. Grids were based on the City of Minneapolis boundaries.

Figure 2: Four Week Hot Spot (Kernel Density) of Part I Crime occurring between Jan, 2011 - Feb 21, 2011.

Many agencies are experimenting with predictive analysis and using a variety of vendors and algorithmic techniques.  With limited resources and a lack of research on the best business solution for forecasting crime, the MPD Crime Analysis Unit continues to test multi-tiered correlation of shapefiles (points, grids and kernel density) built from suspect, victim, loss and location data to substantiate focus zones and predict trends. To create future-oriented analysis, categories and tables in the 20 year old report management system are broken down and imported into ArcGIS shapefile layers, emulating many of the contemporary methods that correlate time, distance, and bearing.

Figure 3: Jan 1, 2010 - Dec 31, 2010 Burglary Grids (Micro Hotspots), Light Blue - 9 to 14 incidents, Dark Blue 15-20.

The Authors

Crime Analyst Ryan Hughes is a graduate of the M.S. Geographic Information Science at St. Mary’s University – Winona Campus, and can be contacted at

Sgt. Jeff Egge is a former research fellow at the Police Executive Research Forum, and can be contacted at

See Also: