Mining Modus-operandi Patterns of Swedish Serial Burglaries

Around 22,000 burglaries are reported to the Swedish police in 2012. It is not only inefficient to analyze these records by human experts, lots of valuable information remains hidden due to weakness of human information processing. Data mining is a promising technique to uncover hidden, unknown and potentially valuable information from large amount of data. The goal of this project is to analyze burglary records and find crime patterns from a burglary dataset by using data mining and machine learning techniques. In this paper from the perspective of data mining I redefine the crime patterns by International Association of Crime Analysts. Then a series of correspondent algorithms and techniques are introduced to mine these patterns. A prototype is implemented to analyze the provided dataset. Crime patterns are identified and visualized in an understandable and user friendly fashion