Chapter Preview
TopThe Use Of Density Analysis In Empirical Studies
In many business and research applications, there is a need to investigate the spatial distribution of some events of interests, in order to identify their geographic “hotspots” and to develop policy remedies, if required (see Table 1).
Table 1. Examples of density analysis use in different scientific fields and applications
| Scientific Field | Research Application |
| Criminology | Forecasting locations of future criminal and terrorist events (Porter, & Reich, 2012) |
| Mapping the addresses of recorded crime events for the identification of crime spots (Wolff, & Asche, 2009) |
| Environmental sciences | Estimating the conditional probabilities of the rainfall (Sharma, 2000) |
| Identification of regions with recurrent forest fires (Gonzalez-Olabarria et al., 2012) |
| Forecasting wind power potential (Taylor, & Jeon, 2012) |
| Epidemiology and public health studies | Identifying the association between disease density and environmental risk factors (Kloog, Haim, & Portnov, 2009; Portnov et al., 2009; Zusman, Dubnov, Barchana, & Portnov, 2012) |
| Analysis of the distribution of alcohol outlets in residential neighborhoods (Carlos, Shi, Sargent,Tanski, & Berke, 2010) |
| Marketing | Analysis of the service area of a restaurant (Donthu, 1991) |
| Modeling the patterns of customer density aimed at locating prospective new customers (Sliwinski, 2002) |
| Transportation | Density analysis of recorded traffic accidents for understanding their spatial patterns (Anderson, 2009; Xie, & Yan, 2008) |
| Space–time analysis of traffic trajectories of passenger ships and tankers (Demšar, & Virrantaus, 2010). |
Key Terms in this Chapter
Kernel Density (KD) Estimation: A density calculation method that weights events that are close to the center of the search circle more than more distant ones.
Raster Cell: Usually a rectangle used for surface tessellation.
Double Kernel Density (DKD): A product of normalization Kernel Density (KD) surface by another KD surface, for instance, by the KD of the total population residing in the study area.
Edge Effect: A KD surface distortion characterized by a drop in the values observed near the edge of the study area.
Linear Averaging: Density estimation method which calculates the number of events inside the search radius divided by the search circle area, without applying any weighting scheme to the input events.
Age-Standardized Rate (ASR): The rate of a disease adjusted to the age distribution of a “standard” population (that is, population of a region or a country, set as a conditional baseline).
Bandwidth: The radius of a search circle used for density calculation.
Kriging: A geostatistical interpolation method based on predicting values for unmeasured locations using regression estimates.