Urban shadows in Latin America: Detection of crime hotspots using multivariate, geospatial, and Neutrosophic Analytic Hierarchy Process (NAHP) analytical techniques

Abstract

In recent years, Guayaquil fell prey to higher levels of organized extortion compared to territorialized criminal organizations over territory, institutional weaknesses, and socio-spatial inequalities. Thus, while extortion emerged as an increasingly specialized crime, there are limited empirical studies to evaluate the crime's spatiotemporal nature or its determinants. Therefore, this study aims to identify and characterize extortion "criminal severity zones" of Guayaquil from 2021-2024, via a novel geostatistical approach, in addition to the Neutrosophic Analytic Hierarchy Process (NAHP). By implementing Exploratory Spatial Data Analysis (ESDA), Kernel Density Estimation (KDE), dimensionality reduction techniques (PCA, MDS, and t-SNE), hierarchical clustering, and Bayesian hierarchical modeling via INLA—as well as NAHP to rank districts for intervention under uncertainty—this research finds statistically significant results. Specifically, using a cleaned, georeferenced variable dataset of extortion-related emergency calls (ECU-911) and socio-demographics from 2022 census data, the study uncovered significant spatial autocorrelation (Global Moran's I = 0.271, p < 0.001), as extortion hotspot high-high areas were found northwest in Nueva Prosperina and northeast in Pascuales. KDE and Getis -Ord G* threshold mapping confirmed extortion hotspots overlapped with poverty hotspots in these areas over various months across time. Clustering via PCA suggested three crime severity zones—consistent/extreme, emerging, and low—based on fluctuations/stability of extortion severity. Bayesian testing confirmed spatiotemporal structures as the posterior relative risk of specific districts exceeded 3.4. Finally, the NAHP revealed extortion prevention priority districts including Guayaquil Island, as data indecision and neutrosophic logic allowed for further justification of risk trends. Such findings are unprecedented as they provide empirical contributions to crime's territorial dynamics of extortion relative to resource-scarce cities in Latin America. Further, applying neutrosophic logic through the NAHP process adds to the rigor of this study while providing a platform for future researchers and decision-makers alike. Ultimately, this paper presents both methodological contributions as well as significant findings that call for evidence-based treatment and policy.