Analyzing the relationship between homicides and police reports in Peru: A multivariate and cluster analysis with a spatial approach

Spatial crime theories explain the geographic concentration of crime and inform understanding of Peru's rising criminal activity. In this regard, the present study analyzes the spatial and statistical relationship between homicides and other crimes reported to the police in Peru, using available data from 2016 to 2021. Using three multivariate techniques, namely, correlation analysis, hierarchical clustering, and principal component analysis, we examine the association between homicides and 30 police-reported crime subtypes. The results highlight a marked temporal and spatial variation at the provincial level. The cluster analysis identifies three main, well-differentiated criminal behaviors. In one of these groups, a strong relationship between homicides and economically motivated crimes is captured, such that provinces with higher homicide levels, especially in Metropolitan Lima and Callao, also show an increase in the proportion of these crime subtypes. In contrast, provinces with lower homicide burden tend to report more non-lethal interpersonal violence crimes. Overall, the findings confirm that homicides are not randomly distributed across the Peruvian territory and, instead, in recent years have become associated with specific crime patterns in certain areas, particularly with economically motivated crimes. In conclusion, the statistical methodological approach employed allows for the identification of differentiated crime patterns and a more precise understanding of the criminal profile at the national and regional levels, offering analytical insights to support the strategic prioritization of criminal offenses and the development of more effective approaches to enhancing citizen security. • Hierarchical clustering analysis revealed three distinct groups of criminal offense patterns across Peruvian provinces. • The use of relative frequencies enables the detection of overrepresented crime types in relation to homicide prevalence. • The absolute number of homicides shows a strong positive correlation with economically driven offenses. • Spatial structuring of provinces reveals that homicide prevalence tends to cluster in coastal and metropolitan zones. • Provinces with low homicide incidence exhibit higher proportions of non-lethal interpersonal violence.