Predictive Crime Modeling Project

The Smart Policing Initiative seeks to build upon the concepts of offender-based and place-based policing and broaden the knowledge of effective policing strategies. The most convincing research demonstrates that place-based or hotspot policing reduces violent crime and neighborhood disorder. This initiative addresses the need for effective policing that requires a tightly focused, collaborative approach that is measurable, based on sound, detailed analysis and includes policies and procedures for accountability. This grant program seeks to build upon data-driven, evidence-based policing by encouraging state and local law enforcement agencies to develop effective, economical, and innovative responses to precipitous or extraordinary increases in crime, or in a type or types of crime within their jurisdictions.

The City of Miami, Miami Police Department (MPD) will utilize Smart Policing Initiative grant funds, in partnership with a research partner, to develop economical and efficient predictive analytics that will improve MPD's ability to make force deployment decisions and increase crime clearance rates, specifically auto-thefts and robberies. The Miami Police Department selected these two areas of improvement due to the following challenges: 1) specific areas of vertical growth in the city have the largest increase in calls for service as compared to all other city neighborhoods that have remained relatively flat, 2) MPD's limited crime spot mapping capability, and 3) MPD's robbery and auto theft clearance rates are lower than the national average.

The MPD will use funds to purchase a predictive modeling software workbench, hire an experienced statistical analyst, who is capable of building and interpreting predictive models, and hire a researcher to conduct an evaluation. The modeler will be capable of providing crime forecasting and criminal behavior analytics in conjunction with, or overlaid against, the existing hot spot technology. The MPD police command staff will use the predictive models to make force deployment or tactical decisions to address areas of concern. The research partner will then analyze and validate the validity of the models with input from experienced crime analysts, in order to quantify the results and to provide feedback on how the models should be adjusted to improve the results further.

CA/NCF