Multi-source Predictive Policing

Project
Conception d'un modèle prédictif et explicable pour la répartition des effectifs de police dans les villes 

Summary
Predictive policing algorithms have gained popularity in recent years, as some have been documented and implemented in major cities. However, these predictive policing solutions are often developed without considering human factors, ergonomics, and the impact of implementation on the work field. Combining design thinking, criminology, and data science, this project aims to design a predictive policing and decision-making tool adapted to public safety situations with high levels of uncertainty. Initially, the real-time pooling of information from multiple sources will provide a better representation of the current situation, priorities, and uncertainties. Subsequently, classification and/or prediction algorithms applied to time series with seasonality will be developed to incorporate the short to medium-term situation prediction component based on historical data and past events. The implemented tool will support managers in resource allocation and patrol assignment to optimize the distribution of police forces.