Contextual Analysis of Crime: A rule-based AI solution for spatial crime risk analysis

The Problem

Prior to the initiative, municipal safety efforts in Edmonton relied on traditional, reactive "hotspot" mapping, which only identified where crimes had already occurred rather than why they happened. This approach left the City unable to anticipate criminal activity or address the underlying environmental, social and physical root causes driving it.

Furthermore, because municipal data was heavily siloed, there was no mechanism to analyze the correlation between localized disorder (such as litter, noise complaints or abandoned vehicles) and actual crime rates. Lacking this centralized, contextual intelligence, operational teams could not optimize their deployment strategies, leading to operational inefficiencies and preventing proactive community intervention.

The Solution

The Contextual Analysis of Crime pioneered the "Crime Lasagna" methodology at the City of Edmonton, which maps traditional crime statistics as a base layer and stacks 233 environmental, social and physical datasets on top. By utilizing Risk Terrain Modeling (RTM) and rule-based machine learning to process over 2.5 million data points across 11,074 localized grids, the system generated 92 distinct predictive rule sets. These rules identify exactly how specific neighborhood conditions statistically alter the likelihood of crime. This granular spatial risk analysis provides operators with highly precise, actionable intelligence to neutralize environmental risk factors before criminal activity can manifest. The Crime Lasagna methodology was built using open source tools within existing budget.

The Impact

The project successfully shifted Edmonton toward a highly efficient, preventative safety model, yielding a $1.60 Social Return on Investment (SROI) for every dollar spent. By replacing reactive policing with predictive, grid-based targeting, the initiative achieved 7.6x more efficient resource deployment and spurred a 38% increase in bylaw compliance to neutralize environmental crime drivers.

Implemented as a pilot in the Boyle Street area, the model successfully reduced property crime, boosted localized resident reporting, and strengthened trust between police, bylaw officers and the community. This innovative framework earned prominent external validation, securing both the 2015 Community i-Performance Award ("Best of Show") and the 2017 Exemplar Award.

Lessons Learned

The project demonstrated that analyzing the surrounding physical and social ecosystem—layering factors like noise complaints, litter and abandoned vehicles—provides the vital context needed to understand why a neighborhood is vulnerable, shifting operations from reactive suppression to proactive mitigation.

Launch year: 2015