This thesis describes research on the use of machine learning techniques in crime analysis and prevention for predicting the overall crime trend in an urban area, as well as the likelihood of crime occurrence in a given local area during a time period.
The author of this thesis demonstrates machine learning techniques that enable higher crime monitoring accuracy than current methods based on local crime density alone. The author uses crime data extracted from Citizen and Law Enforcement Analysis and Reporting (CLEAR) system in Chicago, Illinois, demonstrating that state-of-the-art learning algorithms can achieve improved prediction accuracy over traditional methods, based on time series models. The thesis discusses prediction techniques for determining the likelihood of crime occurrence at a specific local area during a given time window, and reports on the demonstration of those techniques in the operational framework of the Strategic Decision Support Centers (SDSCs) in the Chicago Police Department, where only a maximum of six surveillance cameras in one district can be monitored simultaneously at any given time. It also reports on the application of those prediction techniques to select the cameras that most likely have crime events happening within their viewsheds during a determined time window, and, as a result, maximizing the crime monitoring efficiency.