Authors:

Eleazer D. Hunt

Ezra B. W. Zubrow

Title:

Building Crime Analysis Extensions for ArcView

Abstract:

The use of GIS with its ability to visualize spatial relationships, facilitates crime analysis, allowing law enforcement agencies to analyze incident data in exciting new ways. This paper describes the construction of a crime analysis extension kit for ArcView that goes beyond simple map displays and "electronic pin maps"; rather, it can provide predictive modeling and contains algorithms to enhance proactive policing.

Funded by a Department of Justice grant, a consortium to build the extension kit was developed, partnering Environmental Systems Research Institute, Inc. (Esri); the National Center for Geographic Information Analysis at the State University of New York at Buffalo; the City of Salinas, Ca, Police Department; the City of Los Angeles Police Department Crime Analysis Unit; and the San Bernardino County Sheriff’s Department Crime Analysis Unit. The police departments will assist in the technical design and needs assessment of the crime analysis tools. The NCGIA will research and build the models, algorithms, and statistical measures for each tool. Esri will assemble the application. The crime analysis application will be a stand-alone set of tools that will be installed and operated with Esri’s popular desktop GIS software ArcView. The approach is to build upon existing technology and extend it to law enforcement agencies. The application will facilitate the access to police databases with the aid of a "data browser." The data browser allows each department to identify the data tables and map data and let the browser build the links to the crime analysis application.

The paper will discuss the status of the project, initial research findings, and anticipated capabilities.

Paper Body:

1.0 Introduction

1.1 Social Issues

Increasing crime appears to be a feature of all modern industrialized societies, and no developments in either law or penology can be shown to have had a significant impact on the problem. On the other hand, improvements in police technology, crime analysis, and efficient use of police resources have had a significant effect. The effect of crime on the quality of life should not be measured simply in terms of the actual incidence of crime. The fear of crime affects far more people than are likely to become actual victims and forces them to accept limitations on their freedom of action. (David A. Thomas, Reader in Criminal Justice, University of Cambridge; Fellow of Trinity Hall, Cambridge; author of Principles of Sentencing). Furthermore, one knows that the definition of criminal activity varies over space and time and that crime analysis tools need to take this into account. The trends in the United States have been toward the creation of new criminal offenses. These social changes have acted to increase the rate of criminal activity.

The problem is not simply the fear of crime, but who is committing those crimes. Crime is predominantly the activity of the young, and thus it impacts the social future to a far greater extent than would be expected. In both Britain and the United States, for example, the peak period for involvement in relatively minor property crime is adolescence—from fifteen to twenty-one. For involvement in more serious crimes the peak age is likely to be rather higher, from the late teenage years through the twenties. It appears that a higher incidence of youth-related crime and gang involvement correlates to deprived neighborhoods of large cities, groups with lower educational/occupational attainment, and adherence of members of some minority groups to cultural standards that are in conflict with the general law.

1.2 Institutional Law Enforcement Issues

The efficiency and effectiveness of a police department depend in large degree on the efficient distribution of police resources, rapid response time (e.g., E-911 systems), and systematic analysis of data for determining incident patterns. The vast majority of law enforcement agencies maintain records in tabular form, without the ability to visualize or conduct spatial modeling. The work that Environmental Systems Research Institute, Inc. (Esri), has conducted over the last twenty-seven years has pointed out the importance of maintaining data and conducting analysis along both spatial and temporal axes. Information must be reported, collected, classified, analyzed, processed, and stored. If a criminal has been successful using certain methods and techniques, he tends to repeat the same procedure. Thus, evidence of characteristics can serve to identify patterns and display the patterns using a geographic information system (GIS).

The problems of increased societal violence, gangs, and increased density of urban populations result in greater problems for police departments in solving and responding to crimes, proactively preventing crime, efficiently administering resources, and maintaining officer safety. As a result, there is now the need to provide law enforcement agencies (LEAs) with technology that can rapidly conduct crime analysis, gain insights into the characteristics of beats, and predict classes of criminal activity.

Developing fast and efficient police data management systems is itself not the goal. The goal is to reduce criminal activity and gang violence and to understand the social structure of the community. Thus, it is necessary for LEAs to work with the community and civic leaders. The City of Salinas, California, has recognized this need and has developed a Community Oriented Policing program and implementation guidelines. The concepts the Police Department promote are commitment to public satisfaction, solving chronic problems, empowering the beat officer to be the problem solver, determining the actual needs and desires of the community, working in partnership with the community to solve problems, personalizing police service response via beats, becoming proactive, and becoming less reactive. The GIS application allows each beat officer to see the type and pattern of criminal activity within their area of responsibility and in effect makes "each officer a crime analyst." The result is that the application has made an immediate impact in the manner in which officers reach conclusions, target criminals, and reduce crime incidents within the City.

1.3 Solution

The use of GIS and related geostatistical analysis for crime analysis is just beginning to gain popularity with (LEAs). The lowering costs of hardware and software have allowed LEAs to analyze crime data electronically, using the power of GIS to gain new insights to the data. As such, LEAs have relied on the capabilities inherent within the GIS software to conduct analysis of crime data or have relied on outside consultants to build custom analysis procedures.

This paper discusses the award by the National Institute of Justice (NIJ) of a grant specifically to build a set of GIS crime analysis tools for law enforcement. The grant was awarded to a consortium consisting of Esri; the National Center for Geographic Information Analysis (NCGIA) at the State University of New York at Buffalo; the City of Salinas, California Police Department; the City of Los Angeles Police Department Crime Analysis Unit; and the San Bernardino County Sheriff’s Department Crime Analysis Unit.

The focus of this project is the involvement of LEAs in the designing and research of the crime analysis tools. Besides the members of the consortium, an additional group of ten to twelve LEAs will participate in the development process.

The objective of this project is to produce a set of crime analysis tools that can be provided as an extension kit to ArcView® desktop GIS software. To facilitate its use across disparate data formats and a wide range of law enforcement agency data structures, the extension kit will include a "data browser". A data browser allows the user to tell the application where each piece of data is located and names of the appropriate fields that need to be accessed.

The goal of the project is to provide a set of tools to assist beat officers in understanding incident patterns and potential threats to officer safety, provide law enforcement agencies with predictive measures of incident patterns, and enable all levels of law enforcement to have an effective crime prevention tool. With this application the beat officer becomes a crime analyst.

The solution will be based on Esri's ArcView Version 3.0 software product. As the most recent software release in Esri's line of desktop software, ArcView Version 3.0 will provide a new level of GIS mapping functionality, as well as an extensible robust architecture from both database and application points of view. This architecture will allow the consortium to develop a set of crime analysis applications that may be effectively implemented within community policing organizations nationwide with little customization required.

The importance of providing a prepackaged set of crime analysis tools within ArcView is that it provides law enforcement agencies with an easy-to-use and integrated approach toward using GIS, allows tabular data to be viewed spatially, and can, in the future, work in patrol cars via laptop computers or mobile computer terminals. These tools represent a long-term highly advanced solution to beat officer support efforts in crime analysis and community policing.

2.0 Structure of the Extension

This section illustrates the structure and process of building the extension. Esri, being a software and applications developer, has developed and refined an application development framework that is based on a logical progression of tasking.

Figure 2.1 illustrates the schema of how the extension fits into the existing structure of ArcView software, and the relationship between data sources and how end users access the data. The data can come from a variety of sources including a records management system (RMS), computer-aided dispatch (CAD), and GIS. The structure of these data is not standardized; there currently exist no data standards for law enforcement. As a result each jurisdiction develops and maintains its own RMS, CAD, and GIS. What the crime analysis extension has built into it is a data browser. The data browser is an interactive process that asks each user to identify the location of each source of data. For example, for a street centerline map that will be used for address matching incident locations the browser will ask for the directory within which the file can be found and the names of the fields within the file that contain attributes such as street name, street type, address range, zip code, and so forth. The data browser maintains the location of each data source and associated attributes. There exists a relationship between available data and functionality of the crime analysis extension; this being, if all the data are available, then all the functionality of the extension is enabled; however, if some data are not available, then the corresponding functions that rely on those data are not enabled.

The crime analysis extension provides two basic functions: applies algorithms and automates crime mapping. It operates via a graphic user interface (menus, buttons, and tools). Esri has designed ArcView Version 3.0 to accommodate additional functionality through the addition of extensions. Currently, Esri has developed two extensions: Spatial Analyst and Network Analyst. When installed, each extension modifies the graphic user interface of ArcView Version 3.0 by adding the necessary menus, buttons, or tools to allow the user to operate the extension. Behind each menu, button, or tool is the algorithm or programming code that is processed. One of the main tasks within this grant is to conduct the research necessary to develop the algorithms for spatially based crime analysis.

3.0 The Tools

The development of a set of crime analysis tools has been completed for the City of Salinas Police Department. The crime analysis tools were developed as part of a Department of Justice (DOJ), Office of Community Policing Services Innovative Problem Solving Grant to reduce youth handgun violence. The Police Department decided to develop a set of GIS-based crime analysis tools in order help officers track gangs; locate stolen, recovered, seized, or found firearms; and display the spatial relationship between specific incident types and subjects.

By working closely with the Police Department's Violence Suppression Unit, a detailed needs assessment was conducted to identify the types and functions of tools and the structure and organization of the police database management system. The tools developed for Salinas include the following:

The set of tools is accessable with the use of a graphical user interface and menus. The complete application consists of ArcView Version 2.1/3.0 desktop GIS software and a custom Visual Basic 4.0 interface. A user builds a query using the interface, and the query elements are passed to ArcView for processing and display. Each map that is displayed can be queried to retrieve specific information for each incident (case number, suspect, crime codes, vehicle information).

The process of developing a GIS-based crime analysis application has illustrated several key areas. First, it has demonstrated the need for standardized data collection. The Police Department has evaluated the way information is being recorded on reports and field interview cards and has made modifications to capture data for better utilization. Second, addresses need to be validated in order to minimize mistakes locating an incident on a street map. Third, for tracking gang members, the Field Interview form has become an important source for seeing the movement of individuals spatially and temporally. Fourth, another level of tools could be developed to provide more robust analysis of temporal patterns in crimes, crime density, and predictive measures for specific types of crimes.

3.1 New Tools

From the work completed at Salinas Police Department, a number of valid conclusions have been formulated. First, the use of GIS to provide spatial analysis of police data has proven to be extremely useful and well received. Second, the use of a graphical user interface in order to make the tools officer-friendly without large amounts of training has received praise and high acceptance by officers throughout the Salinas Police Department. Third, the entire process of updating and maintaining data by accessing the police records management system and porting the information for processing in the GIS is documented and understood, thus enabling other departments to gain from this knowledge. Fourth, for the basis of researching and developing new tools, the existing Salinas application does not require a completely new set of data collection or processing—the data already exist in a form for immediate use.

The result of this collaborative work between Esri and Salinas Police Department has created demands for more tools with differing capabilities. A number of different types of tools and functions have been discussed. These include the following:

1. Computerized Beat Book

This tool takes two approaches. The first approach is to facilitate officer knowledge of individual beats (street names, street direction, barriers, and businesses). Police departments stress the need to educate officers about each beat and state that departments are constantly conducting training due to new officers or switching officers to different beats. By computerizing the beats, officers can look at information about each beat or print out maps they can use while on patrol. The second approach is to allow an officer to see incident activities within a beat (e.g., daily activity status of all three patrol shifts, parcel information [permits and licenses]), hazards (previous violent crimes, drive-by shootings, parolees, or gang activities), probability of a crime occurring in an area at a specific time or day (uses previous incident data and crime density to predict potential types of incidents and allow officers to be cautious or reroute patrol pattern), gang activities (level of activity, type of incidents), and pattern of incidents (spatial, temporal, and day of week).

2. Automate Gang Territory Boundaries

This tool automates the building of gang territories and allows law enforcement agencies to monitor the changes in territory boundaries along with gang activities outside of any territory. Territory boundaries can be discerned through a combination of confidential informants, tagging associated with specific gangs (using either addresses or global positioning system [GPS] to locate the position of the tag), gang associated incidents (drive-by or criminal act) that can be attributed to a specific gang, and field interview location (identify or tattoo association). This tool works on the basis of pattern analysis algorithms combined with Fourier transform and predicts where the boundaries will be for short periods into the future. There are good data to support a one- to two-day period prediction based on the past six to eight days. Period subdivision is used to improve predictive results. It shows the dynamic changes of gang boundaries, directional expansions or contractions, and trends.

3. Optimizing Police Officer Resources

This tool is designed to more efficiently allocate patrol resources. There are several different functions within the tool. First, from an administrative and planning perspective it would determine, on the basis of station locations, sizes, response areas, transportation pattern, and type of incidents, what is the optimal distribution of police for maximal coverage (heightened presence in higher incident areas) and spatial optimization (rapid response with routing models). This could use upgraded versions of PCAM or HyperCUB car allocation models. Second, it would display the history of crime patterns during a particular time of day and/or day of week and build a patrol route based on the optimal response time or being in the vicinity of higher density of incident locations. Third, it would display crime patterns by specific type of crime over a longer period (such as the previous week, month, year) and based on probability maps build models of optimal distribution routes for patrolling for specific crime type.

4. Beat Boundary Tool (Optimizing the Beat)

This tool takes the existing DOJ redistricting application and adds functionality by incorporating crime incident data, demographic data, dispatch incident data, natural barriers, routing, and land use in order to create methods/formulas that allow law enforcement agencies to build boundaries in an interactive fashion. The existing application operates in ArcView Version 2.1. This task will focus on developing a customized beat boundary tool in ArcView Version 3.0.

5. Police Administration Reporting Tool

This tool would provide geographically specific data on which special enforcement programs (COPS programs, curfews, gang territories) are most effective by spatial unit (beat, reporting district, or program boundary). One could correlate such programs with types of crimes for each area in order to assess effectiveness before and during the duration of the program.

6. Spatial Choropleth Tool

This tool would produce maps of general crime density and crime density by type of crime, at specified resolutions for specified periods. It would produce contour maps of the density based upon street blocks, census blocks, beats, reporting districts, or user-defined area. The result will be used by several other tools in order to calculate modeling algorithms.

7. Officer Safety Tool

This tool would produce maps or, more importantly, give spatial locations where officer safety is an important issue. This tool would take into account all previous responses to complaints that were met by a violent or potentially violent activity within a particular distance of the dispatch call and suggest the level of violence the officer might expect based upon previous dispatches. When officers are dispatched to a particular location they could be given such information (such as over the last three months, within one block, there have been reported three knifings and one shooting).

8. Social Census Tool

This tool examines a variety of social indicators based on census data to understand the demographic profile of any given area within a municipality. Based on the demographic profile, a series of correlations is conducted against crime incidents. This creates a capability to geographically identify social indicators and activities related to criminal activities such as gangs, guns, and youth activities. It also allows one to identify high probability areas and show the changing trends over time within them. Finally, it allows predictions of future potential incidents by forecasting demographic profiles.

3.2 Tool Discussion

The basic criteria in tool development includes

_ Provide analysis capability of demographic data or utilize demographic data as part of specific tools.

 

_ They should be machine portable (i.e., portable on platforms).

 

_ They should be designed for speed in requesting and retrieving information.

 

_ They should be designed for future use in patrol cars using mobile computer terminals.

 

_ They should be designed for ease of use by line officers/police beat officers for efficient crime analysis.

 

_ There should be built-in safety and security measures.

_ The application set should facilitate setup across a range of data structures with the use of a data browser, allowing users to identify the spatial theme sources and necessary fields for classification and identify attribute tables and necessary fields for links/joins/analysis.

 

_ Build upon ArcView Version 3.0 capabilities and the Spatial Analyst extension.

 

_ Any tool set should be field tested by various types of law enforcement agencies.

 

4.0 Current Status

At the time of this writing the grant has just been awarded and the initiation meeting has not been held. The NCGIA has been conducting research in preparation for the start of the grant.

The NCGIA has finished the preparatory work for the development of GIS algorithms for crime analysis. The process began with an extensive literature review of the spatiotemporal aspects of criminal behavior to familiarize the team with current research. Additionally, the team has examined preliminary efforts, from around the country, to incorporate GIS in crime analysis. Finally, they have made initial contact with numerous police to establish an ongoing and cooperative relationship with law enforcement agencies. These relationships will foster an environment in which the academic and professional will work hand in hand in all stages of the development process.

Specifically, considerable time has been devoted to handling several methodological issues. In particular, individual members of the team have studied statistical techniques for aggregate data analysis, cross-level inference, and contextual analysis. Since crime statistics are kept in the aggregate and it is necessary to make inferences about individual events, these techniques will prove useful in developing the algorithms. Also, the team has investigated more conventional techniques of crime analysis.

Important to the creation of the algorithms is an understanding of the various ways in which law enforcement agencies store and process their crime data. The vast myriad of database structures being used by law enforcement agencies around the globe need to be taken into consideration if a truly flexible and universal GIS can be designed. The algorithms have to be able to incorporate the data in a variety of

structures and be able to distinguish between the differing taxonomies used.

The next set in this process is to use real crime data to design the cluster models and decide which forms of spatial statistics should be used. Once the statistical and mathematical aspects of the project are completed, the NCGIA programmers will then produce the code for the initial algorithms to be used by Esri.

5.0 Conclusions

The immediate benefits of this set of tools is to incorporate spatial analysis and display into crime analysis and beat officer safety and to provide these capabilities in an "off the shelf" manner that can be used by almost any law enforcement agency. The tools themselves are structured in two areas: societal and institutional. In the societal area the tools assist in understanding the demographic characteristics of beats or neighborhoods, possibly create time savings to save lives, reduce crime occurrence, and work to save neighborhoods. Furthermore, the use of predictive modeling results in more proactive policing. For institutional needs of law enforcement agencies, the tools are generalized enough to be modified for specific departmental needs, easily added to a range of different police database management systems, and ultimately support administrative needs of each agency.

Author Information:

Eleazer D. Hunt, Ph.D.

GIS Analyst

Esri

380 New York Street

Redlands, California 92373

Phone: 909-793-2853

Fax: 909-307-3014

Email: lhunt@Esri.com

Ezra B. W. Zubrow, Ph.D.

NCGIA

Department of Anthropology

State University of New York at Buffalo

380 MFAC

Buffalo, New York 14261

Phone: 716-645-2511

Fax: 716-645-2511

Email: zubrow@acsu.buffalo.edu