Recent advances in police open data initiatives across the United States allow for analyses of datasets in criminology and policing across a variety of spatial scales. However, existing statistical methods often do not allow for the utilization of the original spatial granularity or event-level information present in these complex datasets. My dissertation addresses gaps in spatial statistical methodology for analyzing crime and policing data. In particular I (1) propose a new approach to analyze areal crime data through the use of both spatial and social dependence between communities, (2) develop a two-stage approach to study marked point process data with flexible relationships between the location-determination and the mark-determination stages of the point process model, (3) adapt a new shared component model for point process data that allows for flexible characterization of shared spatial patterns between point patterns, and unique drivers of each point process, and (4) explore a method for privatizing data that preserves its statistical utility. (Publisher abstract provided.)
Similar Publications
- DNA-Based Identification of Forensically Important Lucilia (Diptera: Calliphoridae) in the Continental United States
- Targeting youth at risk for gang involvement: Validation of a gang risk assessment to support individualized secondary prevention
- Machine Aided Speaker Identification - Phase One - Semiautomatic Speaker Identification System - Final Report of Phase One Research