V-DS logo Visual Data Science Lab FGV EMAp · Rio de Janeiro

Crime Analytics

Where, when, and why crime happens, and what would reduce it

Crime Analytics develops visual analytics tools to understand recorded crime in Brazilian cities and to support public-safety policy. Mirante, CrimAnalyzer, and CriPAV let analysts explore where and when crimes concentrate, from city-wide hotspots down to street segments. CounterCrime adds counterfactual explanations that show which socioeconomic and urban changes would turn an unsafe region into a safe one. Our recent work models the city as a graph whose nodes carry time series, detects spatiotemporal anomalies in it, and explains them as grounded narratives with large language models (AnomalyExplain, Best Paper at SIBGRAPI 2025; GroundSTory). We also study how signs of disorder such as graffiti relate to crime records, and we synthesize this experience for practitioners in a book chapter on urban intelligence for public safety.

Crime Analytics

Researchers

Publications