U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

On the Application of Fuzzy Clustering for Crime Hot Spot Detection

NCJ Number
214472
Journal
Journal of Quantitative Criminology Volume: 22 Issue: 1 Dated: March 2006 Pages: 77-105
Author(s)
Tony H. Grubesic
Date Published
March 2006
Length
29 pages
Annotation
This paper explores the use of a generalized partitioning method known as fuzzy clustering for hot-spot detection.
Abstract
Results on the use of fuzzy clustering suggest that the geometric properties of convex hulls are useful when combined with the results from partition-based cluster analysis in the delineation of crime hot spots. In addition, fuzzy clustering is a better technique for handling intermediate points and spatial outliers when compared to other non-hierarchical clustering algorithms used for spatial applications. The empirical findings demonstrate that in many applications, fuzzy clustering offers a more realistic approach to the delineation of crime hot spots in urban settings. One of the fundamental challenges in crime mapping and analysis is pattern recognition. In detecting hot spots where crime is elevated, efforts and methods are wide ranging. Cluster analysis, which is a rapidly increasing method for hot spot detection, is a technique that seeks to place individual observations into groups that minimize within-cluster variation and maximize between-cluster variation. This paper provides a brief review of hierarchical and non-hierarchical clustering methods, identifying their strengths and weaknesses when used in spatial data and hot spot detection. The paper also explores the relative utility of a generalized partitioning method known as fuzzy clustering for crime hot spot detection. Instead of observations being assigned to one and only one cluster, the fuzzy approach allows for some vagueness in the data. Lastly, the paper explores methods for visualizing differences in cluster probability surfaces with data generated in the fuzzy clustering approach. Figures, tables, and references