An Introduction to Point Pattern Analysis using CrimeStat Luc Anselin Spatial Analysis Laboratory Department of Agricultural and Consumer Economics University of Illinois, Urbana-Champaign http://sal.agecon.uiuc.edu/ June 24, 2003 Introduction This is a brief introduction to the analysis of patterns in points (as events) using Ned Levine’s CrimeStat 2.0 software package. This package is freely available and can be obtained on the web from http://www.icpsr.umich.edu/NACJD/crimestat.html . The data used in this tutorial are the Pittsburgh homicide locations (various Pitt* files) and the Cardiff juvenile offender addresses (juvenile), both obtainable as shape files from the SAL sample data repository http://sal.agecon.uiuc.edu/stuff/data.html. Some familiarity is assumed with either ArcView or ArcGIS, optionally with the Spatial Analyst extension, to implement visualization of various results. CrimeStat does not have its own visualization capability, but relies on an external GIS through the export of result files. Getting started with CrimeStat Start CrimeStat by double clicking its icon. Click on the welcome screen to open the main interface, shown in Figure 1. Figure 1. CrimeStat opening screen.2Note how in some systems (like the one used for this tutorial, running Windows Xp) the bottom buttons are not fully legible. They stand for, left to right, Compute, Quit and Help. Help brings up an extensive help system. The four tabs at the top of the interface correspond to some logical steps in the way CrimeStat implements analysis. First, one needs to set up the data and possibly set some options, then choose the type of analysis (spatial description or spatial modeling). The analysis is run by clicking on the Compute button in the bottom left. Data setup in CrimeStat CrimeStat reads data from various format files, including shape files. You will be using the juvenile.shp data set from the Bailey-Gatrell text in this example. This is a very simple data file with only the X, Y coordinates of the offender addresses. You can load this into ArcView to have a quick sense of the overall pattern, as in Figure 2.1 Figure 2. Juvenile point pattern in ArcView. In CrimeStat, click on the Data setup tab to bring up the interface, shown in Figure 1. In this case, there is only one point pattern, so you only need to specify the Primary File parameters. For case-control studies, you would specify the cases as the Primary File and the controls as the Secondary File. Many routines in CrimeStat also use a Reference File, which is essentially a rectangular grid superimposed over the data points, in order to carry out density estimation or interpolation. In the Primary File tab, specify the juvenile.shp file as the input file. Click on the Select Files button and on Browse in the File Characteristics dialog. Move around in the file system until you locate the juvenile.shp file in your working directory, as shown in Figure 3. Select this file and confirm the selection in the File Characteristics dialog (click 1 Open ArcView, and, with the Views icon active, click on New. Then “add a theme” by clicking on the + (plus) icon and locate the juvenile.shp file. Make sure to click on the check mark to make the theme visible.3OK) and the file name will appear in all the input fields in the user interface, as shown in Figure 4. Figure 3. Select input file (Primary File) Figure 4. Primary File available for the specification of variables. Several options are available to specify weights, time and directional effects besides the X, Y coordinates. You will not be using these additional items in the current exercise, but feel free to explore their use (CrimeStat comes with an extensive manual and several sample data sets). For now, set X to X and Y to Y. Also make sure the coordinate system is set to Projected. You can ignore the data units and time unit since there is no time in the Juvenile data set. Your Data setup interface should now look as in Figure 5.4 Figure 5. Variables and projection selected. Next, specify a reference grid that will be used in the kernel density estimation routines. Click on the Reference File tab and check the Create Grid radio button. Now, specify 0, 0 as the lower left and 100, 100 as upper right as in Figure 6, and leave the default to 100 grid columns. In the Cardiff data set, the actual bounding rectangle is 2, 6 to 94, 95 (you can find out “manually” using the Identify function in ArcView). Figure 6. Specifying a Grid as a reference file.5Save the Grid specifications for later use to a file in your working directory. This is a two-step process. First, you “Save” the grid specification by clicking on the Save button in the interface (Figure 6). This brings up a dialog to name the particular grid setup, as in Figure 7. However, this does not save it to a file. To store this (and other) named grid specifications in a file, first Load them (Load button in Figure 6) and then Save to File in the following dialog, as in Figure 8. Practice Start a second instance of CrimeStat and use the Pitthom.dbf file as the primary file. Set up the coordinates and the reference grid (use the Identify button in ArcView to determine the coordinates of the lower-left and upper-right corners of the bounding rectangle). Figure 7. Save reference grid. Figure 8. Load and save reference coordinates from/to a file. Centrography Basic descriptive statistics of the overall pattern of the points are the mean and median center and the standard deviational ellipse. These summaries are computed in CrimeStat as part of the Spatial Description tab under the Spatial Distribution items. The output can be saved to a shape file (as well as other formats) for overlay on the point pattern. With the juvenile point pattern as the Primary File (and with the X and Y coordinates specified), click on the Spatial Description tab to bring up the Spatial Distribution dialog. Select the check boxes next to Mean center and standard distance (Mcsd), Standard deviational ellipse (Sde) and Median Center (MdnCntr), as in Figure 9. Also specify file names for the output to be saved to a shape file (make sure to select ArcView “SHP” as the option in the Save output to list, as in Figure 9). This needs to be done for each descriptive statistic. Click on the Compute button (lower left) to start the calculations.6The results will appear in a screen, as in Figure 10. Note the tabs on the