Joint Modeling of Multiple Crimes: A Bayesian Spatial Approach

Abstract

A multivariate Bayesian spatial modeling approach was used to jointly model the counts of two types of crime, i.e., burglary and non-motor vehicle theft, and explore the geographic pattern of crime risks and relevant risk factors. In contrast to the univariate model, which assumes independence across outcomes, the multivariate approach takes into account potential correlations between crimes. Six independent variables are included in the model as potential risk factors. In order to fully present this method, both the multivariate model and its univariate counterpart are examined. We fitted the two models to the data and assessed them using the deviance information criterion. A comparison of the results from the two models indicates that the multivariate model was superior to the univariate model. Our results show that population density and bar density are clearly associated with both burglary and non-motor vehicle theft risks and indicate a close relationship between these two types of crime. The posterior means and 2.5% percentile of type-specific crime risks estimated by the multivariate model were mapped to uncover the geographic patterns. The implications, limitations and future work of the study are discussed in the concluding section.

A Bayesian spatial shared component model for identifying crime-general and crime-specific hotspots

“…Recognizing that the local patterning of many crime types are similar, recent studies have used multivariate models to examine the ways in which multiple crime types are correlated over space and space-time. For example, Liu and Zhu (2017) and Chung and Kim (2018) use Bayesian models featuring the multivariate conditional autoregressive prior distribution to capture the correlations between two and three crime types, respectively. Both of these studies observe that accounting for the between-crime type correlation structures improves model fit compared to separate type-specific analyses.…”

Grid-Based Crime Prediction Using Geographical Features

“…this study proposes combining time and space features along with new geographic features obtained from the Google Place API to improve predictive performance. In terms of algorithms, various machine-learning models, such as Naïve Bayes [15,16], Ensemble [17], or Deep Learning Structure [18] have been used for crime prediction, but Deep Neural Networks (DNN) provided better results in our previous experiments. This study uses DNN because it reflects representation learning and has been used in crosslingual transfer [19], speech recognition [20][21][22][23], image recognition [24][25][26][27], sentiment analysis [28][29][30][31][32], and biomedical [33].…”

Anisotropic Diffusion for Improved Crime Prediction in Urban China

“…Marie et al, found that a decreasing rate of undereducated people led to a significant decrease of crime rates against property in England and Wales [28]. Still, few studies have been focused on the relationship between population and crime rate, despite household and population density often appearing as independent variables with significant correlation to crime rates [29][30][31][32][33][34][35]. Law et al, used Bayesian spatial regression to analyze the correlation between economy, population density, ethnicity and crime rate in Yorkshire, Canada aiming to address juvenile delinquency [18].…”

“…Law et al, used Bayesian spatial regression to analyze the correlation between economy, population density, ethnicity and crime rate in Yorkshire, Canada aiming to address juvenile delinquency [18]. Likewise, Liu and Zhu found a significant correlation between burglary rate and household density in Wuhan City, China based on Bayesian regression [34,35].…”