Content area
Abstract
The analysis of crime events has advanced beyond simple pin maps. It incorporates both basic spatial statistics and more complex computational methods for pattern and density discovery. Unfortunately, visualization has lagged behind algorithmic methods and techniques. This has contributed to a lack of standard procedures for data visualization in criminology and crime analysis (Maltz 2014, 5581). The addition of a temporal component adds to the complication, as it continues to create difficulty with the interpretation of space-time phenomena. When comparing crime across different regions, visualization is further confounded by the different methods in which data is collected and maintained by the respective local government agencies, like the police and city planning departments. There are several known attractors of crime, like drug markets, bus stops (Block and Block 1995; Brantingham and Brantingham 1995; Weisburd and Green 1995; Hart and Miethe 2014), and specific land use concentrations (e.g. bars, motels and public housing). Current cluster visualization methods (e.g. point density and kernel density estimation) are limited in their approach to revealing hidden attractors of crime across space and time. The purpose of this study is to develop computational, multi-dimension geospatial and temporal attribute data models to discover hidden crime attractors in institutionalized, high-density cluster locations — meaning areas where a high-density of crime events has persistently occurred over ten or more years. Crime data from 2004-2013 collected from three police jurisdictions were combined with non-police data (e.g. census and transportation) to answer the following questions: (1) What geographic factors are highly correlated with reported crime and are these factors spatially and temporally similar each year? (2) Which locations of persistent high-density crime cannot be explained by the factors revealed in Q1; and, what types of mapping and visual analytic methods can be used to discover additional factors? (3) How can the factor-based methods used to answer Q2 be used to explain spatiotemporal patterns of crimes both over time and across/within the three cities? This work was expected to enhance crime analysis methodology by testing and validating data analytic methods for temporal and attribute visualization from a geographic perspective, thus informing knowledge discovery.
