Characterising Crime Incidence in Mexico: A Hierarchical Linkage Clustering Analysis

Abstract

We aim to characterise the incidence and distribution of crime in 32 states of Mexico. A hierarchical grouping with average linkage is implemented with crime information from DataMéxico, segmented by state and type of crime from 2015 to 2020. Based on the proportional number of crimes over the population of each state, through the elbow method and the average linkage with a Cophenetic correlation coefficient, we validate the number of clusters. Subsequently, a principal component analysis (PCA) is performed to identify each state’s contribution to the clusters proposed. The main results reveal criminal activity can be characterised by three groups. Drug trafficking is the crime that leads the first group, which in turn generates subgroups of interrelated crimes, such as crimes against the family, sexual abuse and harassment, and falsehood, to name a few. These crimes are committed homogeneously in most of the states of the country. Correspondingly, domestic violence and theft lead clusters two and three, and present significant concentration levels since four states accumulate 62% and 55% of crime incidence respectively. The results also provide an overview of how a particular crime can trigger the presence of others.