CMU's CityScan Powers Rodent Prevention and Crime Prediction in Chicago

Dec 26, 2013

Image associated with the CMU's CityScan Powers Rodent Prevention and Crime Prediction in Chicago news item

Funded by the Bloomberg Mayor's Challenge, Carnegie Mellon University and the City of Chicago's Department of Innovation and Technology (DoIT) are currently collaborating to build the "Chicago SmartData Platform", an open-source predictive analytics platform that will help city leaders make smarter, faster decisions in real-time to help address and prevent problems before they develop. By incorporating predictive analytics into the day-to-day operations of over a dozen city departments, CMU will help the city to improve delivery of core services and to effectively respond to a range of issues (including public health, sanitation, traffic management, public safety, and infrastructure maintenance) before they become full-fledged crises, saving money, time, and lives.

CMU's collaboration with the City of Chicago started five years ago, when the Event and Pattern Detection Laboratory (EPD Lab), directed by Heinz faculty member Daniel Neill, began working with the Predictive Analytics Group at the Chicago Police Department (CPD). Prof. Neill and his Heinz faculty colleague Prof. Wilpen Gorr developed CrimeScan, a software package based on novel machine learning methods for event detection and prediction. This tool enables advance prediction of emerging hot-spots of violent crime with high accuracy at very fine spatial and temporal resolutions, and it was incorporated into the CPD's day-to-day policing operations for crime prediction and prevention through targeted patrols.

While the CrimeScan methodology and software have continued to evolve through CMU’s work with the CPD, the establishment of the City of Chicago's Department of Innovation and Technology gave us an opportunity to apply these methods to many other urban challenges through a broader collaboration with the City. The EPD Lab generalized the CrimeScan approach to create the "CityScan" software, which now forms the analytical core of Chicago's SmartData platform, enabling incorporation of many other data sources and prediction of many event types relevant to city operations.

In collaboration with DoIT, the EPD Lab is currently using CityScan to predict emerging clusters of 311 calls for service, enabling the city to anticipate citizen needs and respond proactively. One of the initial focus areas for 311 call prediction is in collaboration with the City's Department of Streets and Sanitation (DSS), focusing on predicting and preventing rodent complaints.

“Using data in a smarter way allows us to improve City services and increase efficiencies, without adding costs for taxpayers,” said Mayor Rahm Emanuel. “This pilot program has the promise of improving rodent services, but it is also the first step in changing how Chicago uses data to enhance the quality of life of its residents.”

Analysis has determined that a combination of requests not directly related to rodents – such as stray animal calls, vacant and abandoned buildings or restaurant complaints – made within seven days in the same general area often result in a service request to eliminate rodents.  By proactively mining this data for these requests every day, the program can deploy a DSS crew to bait before rodents can take advantage of conditions.  The rodent baiting pilot is the first benchmark in the City’s development of the SmartData Platform. Once completed, the SmartData Platform will provide leaders the ability to analyze millions of lines of data in real-time, helping officials make smarter, earlier decisions to address a wide range of urban challenges. 

“By predicting where clusters of these complaints are likely to occur in the near future, we can more precisely target the City's preventative baiting crews, with the goal of reducing the overall level of rat infestation,” said Daniel Neill.

“We are currently working with DoIT and DSS to conduct a randomized, controlled experiment to determine whether we can reduce the number of rodent complaints by predicting, targeting, and preventing rat infestations before they occur. If our experiment is successful, it will not only lead to fundamental changes in the way that Chicago predicts and prevents rat infestations, but will also demonstrate the potential value of CityScan for many other urban issues that can benefit from application of our cutting-edge predictive analytics approaches.”


Interested in supporting Heinz College students and initiatives like those featured in this story? Click here for more information.