Method To Assess, Classify And Project Theft Clusters In Colombia By Means Of Clustering And Neural Networks

Abstract

The purpose of this research is to establish theft clusters in Colombia in order to assess, classify and project the profiles of violence in each region by means of clustering and neural networks. The theoretical framework is based on the articulation of artificial intelligence, neural networks, data analytics, clustering and theft in Colombia. In the methodology, a cross-sectional study was conducted based on a quantitative analysis, starting from historical data generated by the National Police of Colombia, on thefts that occurred in the 32 departments of Colombia in the year 2021. As a result, three theft clusters were established, which were used to identify the region’s most and least affected by this problem. Subsequently, a double-layer neural network was proposed that allowed forecasting with an accuracy level of 96.97% the belonging of each context to a specific theft cluster that will exist in Colombia in the future.