Civil security in urban spaces: Adaptation of the KDE-Method for optimized hotspot-analysis for emergency and rescue services

Abstract

This article gives insight into a running PhD-project. The focus lies on the adaption of kernel density estimation to optimize hotspot analysis for big geodata in the context of civil security research in urban areas. The parameters of kernel density estimation are commonly set in an exclusively mathematical way or using rules of thumb. The element’s spatial component is then left disregarded. This causes an enormous risk for geovisualization in the form of hotspot maps. Urban areas may be declared as police or emergency hotspots despite there only being a small or even not significant random sample of event data. Furthermore there is the risk of over-or under-smoothing of real, existing hotspots in the visual output because of a too small or large cell size for the map grid. That may lead to incorrect tactical, strategic, and operational planning for agencies and organizations with civil security tasks. The method chain presented here offers a relative simple and semi-automatized solution of this problem.