Transportation Research Board 91st Annual MeetingTransportation Research Board · 2012 · 19 citations · 14 references
PlanningCriminologyCrimeSpatial AnalysisSpatial StatisticsCrime Pattern AnalysisCrime PreventionDistribution AnalysisHot Spot AnalysisKernel Density EstimationSpatial Data AnalyticsCrime Hot Spot AnalysisHot Spot IdentificationHotspot AnalysisSpatial Analysis Of Crashes
A number of studies have shown different results in identifying crash hot-spots using different methods such as repeatability analysis, Moran I, Getis-Ord (Gi*), and Kernel Density Estimation (KDE). However, few researchers have discussed the reasons explaining the above differences; nor have they provided guidelines for selecting the most appropriate method for identifying hot-spots. Moreover, only a limited number of studies have analyzed crash data in conjunction with other types of data, such as those related to crime. In this research, a strategy is presented for choosing the most appropriate analysis tool to identify hot-spots that are consistent with the study’s limitations, data characteristics, and objectives. This study used crash and crime data provided by the College Station Police Department. Two important issues were addressed: (1) How to apply data characteristics to the problem of hot-spot identification (e.g., network restrictions for crashes, different hot-spot definitions for crimes and crashes); and, (2) How to incorporate spatial-time distributions in order to identify hot-spots. First, the solution was to apply network distance instead of straight-line distance as the point distance for any calculations, set a spatial limit of KDE, and combined KDE and Gi* maps to capture high risk areas and cluster patterns. Second, a spider graph was used to identify the temporal patterns of events occurring at different time scales: daily and weekly. Then, a co-map that combined a series of KDE maps according to the time of events was used to illustrate the spatial-temporal relationship of crash data to other events. The framework of this study can be used to identify high-risk areas for crash hot-spots and crime events.
14