Uso da análise de cluster para o estudo da criminalidade no Estado do Rio de Janeiro

This article aims to construct clusters based on historical data of thefts in the State of Rio de Janeiro, aiming to identify possible similarities among the records. Monthly quantities of vehicle thefts, robberies on public transportation, pedestrian robberies, cell phone thefts, cargo thefts, and robberies at commercial establishments were selected. Using these records, the k-means algorithm was employed to build clusters, resulting in two subsets of records. These subsets present distinct characteristics and are valuable for analyzing the interaction between different types of thefts in a disaggregated manner, thus avoiding statistical fallacies. Additionally, we propose a classification model that establishes criteria for assigning scenarios to a specific cluster. This model can assist in developing more effective strategies in public security, and in the use of human and logistical resources.