This week's lab covered spatial and temporal crime analysis by comparing three different methods of determining crime hotspots. The map layout here shows the kernel density method, grid analysis, and local Moran I method. They are based on distribution of points representing locations of burglaries.
To use the kernel density tool, we used the point data for burglaries to make contour map and showed the hotspot with the values at least 3 times the mean density. With the grid-based thematic mapping, burglary point data is overlaid on a grid and the occurrence of burglaries within each grid cell is shown. The top 20% or quintile of the grids are used to create a single polygon which represents the hotspot. The local Moran's I method is a little different, because can be thought of as having a predictive quality. Rather than only representing the current year's burglary point density, it also shows clusters of higher than normal density. Areas with slightly lower crime densities that would not show up on the kernel or grid map hotspots do show up on this map because of their proximity to high crime areas. This goes along with criminology near repeat theories that state that crimes are much more likely to occur where they have occurred in the past, as well as near those areas. Examples in the above map are the pink areas that do not intersect the blue (grid) and yellow (kernel density) hotspots. These are areas that police should monitor a little more carefully, because of their proximity to high-crime areas, and the tendency of crime to spread spatially.
The Kernel and local Moran I analyses were carried out using tools in the Spatial Analyst toolbox: Kernel Density and Cluster and Outlier Analysis tools. The Grid map was accomplished using the Spatial Join tool in the Analysis Overlay toolbox. Point data for burglary locations in 2007 and 2008.
No comments:
Post a Comment