Analysis of Mapping Techniques on a Spatial Scan Statistic

The discovery of anomalous regions within spatially oriented data can provide valuable insights to the study of disease prevention, crime, and socioeconomic inequality. Spatial scan statistic methods are a common tool used within these fields to discover areas of a data set that have significant variation when compared to the background distribution. However, the effects of data representation on these techniques has not been rigorously studied. In particular, the effects of point-to-region and region-to-point mappings on scan statistics have not been explored in detail. This thesis will attempt to shed light on the effects of mapping techniques on county and zip code region sets. Data values associated with regions will include cancer rates, educational attainment, and poverty. Additionally, a synthetic data set will be used in an attempt to evaluate the performance of these mapping techniques.