Trend Detection in Crime-Related Time Series with Change Point Detection Methods

Abstract

Time series analysis can be an asset in the hands of the authorities, as it can enable the understanding and monitoring of trends of criminal activities. In this work, a variety of methods is exploited to detect significant points of change in crime-related time series that may indicate the occurrence of events that require attention. In particular, change point analysis is applied in relevant time series, both offline (retrospective change detection when all data is available) and online (detection of changes as soon as they occur). The focus is on the Crimes in Boston and London Police Records datasets, examining how change point detection can benefit relevant authorities in understanding crime trends to better allocate and manage resources. The experimental results allow us to gain valuable insights, including the observation of seasonal patterns in some cases, with corresponding crimes peaking at specific times, the somewhat different change points identified by online and offline methods, and the observation that domain knowledge is desired for better method selection and parameters configuration.