In order to find ways to exploit historical data to generate a better understanding of the dynamics of crime (homicides) in Bogota, we implemented a novel predictive model.
In this paper we have introduced a novel methodology for predicting low-frequency crime events such as homicides, using information obtained from a graph structure, which is constructed with spatio-temporal data. The proposed methodology is largely based on spectral filtering from graph signal processing theory, which allows the precise definition of a Laplacian or Gaussian boundary detection filter operating on graph domains.
The results derived from the proposed methodology were presented to the Secretariat of Security, Coexistence and Justice, and were as follows:
In Projects, you can learn about real applications of our services, use cases and examples of them.