Using a classification model to proper deploy police patrol to face bank robbery in northeast Brazil

Abstract

This article proposes a new approach to help police officers fight bank robberies, espe-cially a violent type of crime named, in Brazil, “Novo Cangaço.” Bank robber y is a massive problem in small towns all over Brazil and, particularly, in the Northeast. In this context, police managers face a complex challenge in deploying their patrols to cover huge areas. To cope with this problem, we propose a new approach using classification algorithms that use the probability of a bank robber y event based on territorial characteristics to deploy more police officers efficiently. We will also analyze geographical features to understand our model, explaining how they impact bank robberies events, using feature importance functions.