Effect of land use on crime considering exposure and accessibility. (July 2019)

• Record Type: Journal Article
• Authors: Sadeek, Soumik Nafis; Minhuz Uddin Ahmed, Abu Jar Md.; Hossain, Moinul; Hanaoka, Shinya
• Is Part Of: Habitat international. Volume 89(2019)
• Publication Date: 2019-07
• Subjects: Crime -- Land use -- Accessibility -- GIS -- Logistic regression -- Support vector machine; Human settlements -- Periodicals; 307
• DOI: 10.1016/j.habitatint.2019.102003
• Languages: English
• ISSNs: 0197-3975

Abstract

Abstract: A substantial number of studies have revealed an association between crime and land use, in some cases by representing land use using socio-economic and demographic data to evaluate their correlation with various types of crimes. The spatial autocorrelation between land use and crime is also often investigated. Many studies focus on data obtained from locations at which a crime has taken place and ignore exposure, i.e., locations where a crime has not taken place. It has also been suggested that transportation accessibility plays a significant role in connecting crime patterns with land use, although this hypothesis requires further study. This paper proposes a new framework to aggregate and synthesize the existing literature based on a geocoding of the association between crimes and land use on a GIS map. The map is broken down into different mesh sizes indicating the occurrence or nonoccurrence of crime within individual mesh cells. An optimal mesh size to best explain the interrelationship between crime and land use with respect to road network accessibility is then developed using logistic regression (LR), and a support vector machine (SVM) is used to identify combinations of land use that are highly susceptible to crime and those that are quite safe. The results of this study can provide insight into reducing crime in cities, allocating law enforcement agency resources, and designing a built environment that can naturally deter crime. Highlights: The study explores association of land use, crime and transport accessibility. It geocodes crimes, land use and road networks and divides study area into meshes. It identifies the best mesh to separate between crime and no crime situation. It identifies the most crime prone and crime deterrent land use combinations. It uses various GIS functions, Logistic Regression and Support Vector Machine.