A Statistical Approach to Crime Linkage

Abstract

The object of this paper is to develop a statistical approach to criminal linkage analysis that discovers and groups crime events that share a common offender and prioritizes suspects for further investigation. Bayes factors are used to describe the strength of evidence that two crimes are linked. Using concepts from agglomerative hierarchical clustering, the Bayes factors for crime pairs are combined to provide similarity measures for comparing two crime series. This facilitates crime series clustering, crime series identification, and suspect prioritization. The ability of our models to make correct linkages and predictions is demonstrated under a variety of real-world scenarios with a large number of solved and unsolved breaking and entering crimes. For example, a naïve Bayes model for pairwise case linkage can identify 82% of actual linkages with a 5% false positive rate. For crime series identification, 77%-89% of the additional crimes in a crime series can be identified from a ranked list of 50 incidents.

Linking property crime using offender crime scene behaviour: A comparison of methods

“…In general, these models achieved statistically larger AUCs than the tree‐based models, which is perhaps unsurprising given the problems of overfitting that are sometimes associated with classification tree models (e.g., Liu et al, ; Thomas et al, ; Tonkin, Woodhams, et al, ). Although, it is important to note that regression and probabilistic models do not always outperform tree‐based models in terms of discrimination performance (Porter, ).…”

“…Within this research, several different statistical methods have been tested to determine whether they have the potential to underpin BCL decision-support tools. These include logistic regression (e.g., Bennell & Jones, 2005;Burrell, Bull, & Bond, 2012;Melnyk, Bennell, Gauthier, & Gauthier, 2011;Tonkin, Grant, & Bond, 2008;Woodhams & Labuschagne, 2012;Woodhams & Toye, 2007), classification tree analysis (e.g., Tonkin, Woodhams, Bull, Bond, & Santtila, 2012), and Bayesian analysis (e.g., Kringen, 2014;Porter, 2014;Salo et al, 2013;Winter et al, 2013). Many of these studies have, however, only tested a single statistical method and have not compared different approaches.…”

“…The regression‐based models did not experience such problems. Other studies have compared regression, classification tree, and Bayesian‐based analysis using samples of sexual assault (Tonkin et al, ) and breaking and entering offences (Porter, ). Tonkin et al () found that a Bayesian‐based approach led to the highest level of predictive accuracy, followed by classification tree analysis and logistic regression, whereas Porter () found comparable levels of accuracy when comparing these three types of statistical approach.…”

Breaking the chain: How arrests reduce the probability of near repeat crimes

“…One way to do this is to identify cases in which there is a higher probability of a near repeat crime (Garnier et al, 2018;Moreto et al, 2014;Piza & Carter, 2018), and then devote more detective resources to those events (or prevent them from being triaged). Another may be to use other automated detection processes to identify if a particular crime is linked to other crimes (Chohlas-Wood & Levine, 2019;Porter, 2016). Again, those criminal events within a linked chain should then get higher priority from detectives, as those individuals have established a prior history that suggests they will commit a future crime.…”

Perfil criminal geográfico: novas perspectivas comportamentais para investigação de crimes violentos no Brasil

“…A análise comparativa de casos é uma técnica utilizada para agrupar crimes relacionados a um mesmo ofensor, ela é realizada a partir da análise de semelhanças comportamentais, das proximidades geográfica e temporais desses eventos [25][26].…”