CrimeStat IV

Susan C. Smith

Christopher W. Bruce

Revised by: Thomas Mueller

About CrimeStat PowerPoint

• This PowerPoint was revised from the Susan C. Smith and Christopher Bruce’s work. The Original can be found at:
• http://www.icpsr.umich.edu/CrimeStat/workbook.html

​

• Their PowerPoint and Manual was developed for CrimeStat 3.0. This PowerPoints was revised for CrimeStat 4.0
• The GeoTech Center wants to thank both of them for allowing us to use this tremendous resource.

​

​

Chapter Five��Hot Spot Analysis�Part 2

Nearest Neighbor Hierarchical Spatial Clustering

• Builds on NNA (NNA determines if a particular crime was more clustered than might be expected by random chance)
• NNH takes the analysis to the next level by actually identifying those clusters
• CrimeStat clusters groups of pairs that are unusually close together
• It creates “first order”, “second order” etc. clusters
• Continues until it cannot locate any more clusters
• Creates both s/d ellipses & convex hulls

Nearest Neighbor Hierarchical Spatial Clustering

• Options that can be used when running NNH
• Fixed distance vs. threshold distance
• Becomes a subjective measure vs. probability
• Minimum points per cluster
• Default is 10
• Alter depending on volume & type of crime
• Search Radius Bar
• Adjust threshold distance and associated probability
• Left – smallest distance, but 99.999% confidence
• Right – greatest distances, but only .1% confidence

Nearest Neighbor Hierarchical Spatial Clustering

• Options (con’t)
• Number of standard deviations for the ellipses
• Single s/d is the default/norm
• Can make small ellipses that are hard to view at a small scale
• Another option is two s/d’s
• May exaggerate the size of the hot spot
• Convex hull vs. ellipse
• Convex hull has greater accuracy
• Convex hull has a higher density than an ellipse
• Convex hulls are defined by the data

Nearest Neighbor Hierarchical Spatial Clustering

• Data Setup; Measurement parameters
• Area = 88.19 sq. miles
• Street Network = 1283.61 miles
• Choose an indirect distance measure.

​

Nearest Neighbor Hierarchical Spatial Clustering

• Spatial description, Hot Spot Analysis I, uncheck Fuzzy Mode, check NNH
• Adjust minimum number points & size of ellipses
• Leave it at random NN distance,
• Change the minimum number of points per cluster to 8. Adjust the size of the ellipses to 1.5 standard deviations.

Nearest Neighbor Hierarchical Spatial Clustering

Nearest Neighbor Hierarchical Spatial Clustering

• In Save ellipses to….Choose SHP, browse to c:\CrimeStat\Data\Lincoln
• Name the file: LFA. Then Click OK
• Save convex hulls to….Choose SHP, browse to c:\CrimeStat\Data\Lincoln
• Name the file: LFA. Then Click OK

​

Nearest Neighbor Hierarchical Spatial Clustering

• Add the projection to each of those files as illustrated in previous labs.
• CNNH1LFA.SHP, CNNH2LFA.SHP (Convex Hulls)
• NNH1LFA.SHP, NNH1LFA.SHP (Ellipses)
• Add to ArcPro GIS, Project, and evaluate

Nearest Neighbor Hierarchical Spatial Clustering

Nearest Neighbor Hierarchical Spatial Clustering

• Experiment with other NNH settings (Example)
• Some of the students set the Points per cluster to 5 and others to 12
• Evaluate the results

​

Spatial and Temporal Analysis of Crime (STAC) – Not covered in this course

• Originally a separate program; integrated into CrimeStat in Version 2
• Produces ellipses and convex hulls
• STAC’s algorithm scans the data by overlaying a grid on the study area and applying a search circle to each node of the grid
• Size is specified by user
• Routine counts the number of points in each circle to identify the densest clusters

Spatial and Temporal Analysis of Crime (STAC)

• You can skip this portion of the lab

Final Notes on Hot Spot Identification

• Clusters are identified based on volume, not risk
• Two areas of town
• 3 burglaries in rural area vs. 20 burglaries in midtown
• Technique to normalize hot spots available
• Risk-Adjusted Nearest Neighbor Hierarchical Spatial Clustering (RNNH)
• Relies on a secondary file with a denominator
• Number of houses, parking spots, etc
• In all of these routines, subjectivity plays a role