Multidimensional Geovisualization of Longitudinal Crime Patterns

• In Event: Innovations in Spatial Analysis

Abstract

Analyzing complex multidimensional crime patterns from a strategic perspective requires the use of several years of various types of structured and unstructured data. Given the uncertainty of spatial data, general exploratory methods utilized for the discovery of unknown attractor-generator attributes are many times insufficient in revealing indirect correlations within long-term high-density crime locations in urbanized areas. This study takes a step beyond the Geographically Weighted Regression method, using self-organized maps (SOM) for further exploratory analysis to expose both known and unknown observable and latent patterns. Traditional mapping techniques along with SOMs and imagery exploitation were used for visualization of high-density crime patterns across time and space to ascertain regional similarities between Atlanta, Georgia and Chicago, Illinois. The crime types examined were street robberies and residential burglaries. It was discovered that both cities, while regionally different, shared noteworthy similarities.

Author

• Crystal English, San Diego State University