Abstract:

The intention of this study was to define clusters of thefts in Colombia, to measure, organize and project the clusters in each context by Machine Learning. This research is based on the integration of concepts from different Machine Learning tools, data analytics and theft in Colombia. As a method, a longitudinal study was carried out, supported by a quantitative rational analysis, with primary information of thefts generated by the National Police of Colombia from 32 departments of Colombia. As a result, the clusters and their prognosis were established with the metrics of membership and accuracy of the Machine Learning models proposed for this research. As future research, the scientific community is invited to replicate the proposed method with other Machine Learning models to evaluate and forecast administrative situations and social problems that can be replicated internationally.