Visualization of Space-Time Dynamics in Criminal Activity

• Activity Number: 216
• Type: Invited
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Sergio Rey*+ and Xun Li and Luc Anselin
• Companies: Arizona State University and GeoDa Center for Geospatial Analysis and Computation School of Geographical Sciences and U and Arizona State University
• Keywords: spatial analysis ; spatial statistics ; open source software ; space-time ; criminology ; statistical graphics

Abstract

In this paper we introduce the package CAST: Crime Analytics for Space and Time. CAST is an open source application based on PySAL (Python Spatial Analysis Library) that combines a suite of new space-time analytic measures together with fully interactive and dynamic geospatial and geotemporal statistical graphics. The new methods include: 1) A LISA Markov chain, which is a space-time transition measure to analyze hotspot growth and the joint transition of an area and its neighbors between Local Indicators of Spatial Association (LISA) categories (hot spots, cold spots and spatial outliers), 2) a conditional spatial Markov chain to assess the probability that a location's crime in a future period is related to the prevalence of crime in neighboring locations in an initial period, and 3) an extension to test for the joint evolution of crime in a location and its neighbors. We illustrate select CAST functionality using crime data for the cities of Mesa and Tempe, AZ.