
I will summarize part of a three year research project on the mathematical modeling of crime in Bogotá, Colombia. This research project was financed by the Colombian Government and conducted by the National University of Colombia, the Secretary of Security of Bogotá and the private firm Quantil. I will briefly describe the main results and techniques regarding crime prediction, interpretability and discrimination or algorithmic bias. Then, I will dive deeper into two of the most challenging problems we faced: the under-reporting of events and the strategic reaction of criminals.