This article was originally published by Data-Smart City Solutions.
Governments today operate in an increasingly complex world, reflected in the volume and ubiquity of data produced by citizens and agencies, as well as the computing power to analyze it. In order to better understand and respond to citizens’ needs and allocate public resources more efficiently, governments must use predictive analytics to leverage this data and develop innovative solutions to contemporary urban challenges.
Predictive analytics is the use of historical data to look for patterns and identify trends, which can be used to reorganize service delivery, anticipate future needs and prevent potential problems. The Inter-American Development Bank (IDB) is working to accelerate the uptake of predictive analytics by governments in Latin America and the Caribbean, spurred by success stories in Montevideo, which is using crime data to analyze patterns of criminal behavior to identify critical points of crime and better target police efforts, and Mexico, which is using data on electricity consumption to get near real-time forecasts of economic activity to inform policymaking. Part of this effort includes launching the “Innovations in Public Service Delivery” discussion paper series, which provides a framework, toolkit and roadmap for public sector employees looking to implement data-driven initiatives. Susan Crawford, Benjamin Weinryb Grohsgal and I contributed the fourth installment of this article series, focusing on challenges, successful case studies and next steps for public officials using predictive analytics.
A number of non-technical, structural obstacles stand between governments and the ability to leverage data for prediction and problem solving. In order to overcome these barriers, governments must:
The discussion paper analyzes in detail two case studies:
City of Chicago: Chicago piloted a predictive analytics project to enhance operational outcomes, specifically in the area of better-targeted rat baiting. By mining the city’s 311 data, a team from the Department of Innovation and Technology (DoIT) observed a relationship among 311 call types that historically correlated with rat infestation problems. The DoIT team took this predictive model to the Department of Streets and Sanitation (DSS) and mapped specific areas for intervention, some of which the DSS had not planned to target. Following the model, the DSS discovered the largest infestation it had ever seen. Thanks to analytics, the agencies were more transparent about what they were doing, resulting in an effective collaboration to resolve a major city problem.
State of Indiana: Indiana launched a Management and Performance Hub (MPH) that facilitates the use of analytics across state databases. In order to better understand the correlation between infant mortality and indicators like health care, nutrition and housing, MPH integrated government data from family, health, finance, business and employment agencies to reveal more obscure trends. Caseworkers could compare a family’s information to past and present data on at-risk families, making a risk estimate that helps determine the probability that a child will be harmed in the future. In this case, the results of the analytics project empowered caseworkers to make data-driven decisions in their efforts to address child welfare issues.
In addition to the Chicago and Indiana case studies recapped above, governments have continued to successfully implement predictive analytics projects in the following key policy areas:
Public Safety: New Orleans worked to identify data and develop a model for a calculated assessment of fire risk to be more proactive and targeted with door-to-door smoke alarm installations. This solution was then scaled nationwide with the help of private and nonprofit partners.
Social Services and Public Safety: Boston and NYC turned to predictive analytics to preempt overcrowding and its consequences, as well as provide affordable, safe housing. Through collaboration across agencies to gather housing data and smart analysis, the cities directed preemptive actions, programs and resources more efficiently.
Public Health: Chicago used advanced data analytics to enhance the process by which health inspectors identify food establishments likely to have critical violations. The city processed open data to find variables that predict violations, developed a model, ran a simulation and used the forecast to allocate inspections more efficiently. Chicago also piloted an analytical model to improve its beach water quality inspection process. A team of volunteers from the city’s civic tech community broadened the scope of the prior predictive model to include new variables and more data, showcasing a successful partner-driven model for municipal analytics.
For more success stories, Data-Smart City Solutions compiled a catalog of civic data use cases that serves as an ongoing, regularly-updated resource for anyone interested in how data and analytics are being used to bolster city operations. The catalog hopes to inspire additional experiments in predictive analytics by cities.
For someone in a Chief Data Officer (CDO) function in government responsible for executing a predictive analytics agenda, there are a number of issues he or she should keep in mind when implementing new projects:
On that last point, we are happy to share that since the paper was published, the Ash Center for Democratic Governance and Innovation at the Harvard Kennedy School has established the Civic Analytics Network, a national peer network of CDOs that collaborate on shared projects that advance the use of predictive analytics and data visualization in tackling urban challenges related to equity.
Citizens today demand better, cheaper, and faster government at both the state and local levels. Public sector employees can be more efficient and more responsive to citizen needs in the context of limited government resources by using predictive analytics to harness the massive volume of available data,
An important caveat is that predictive analytics should be used to enhance the effectiveness and discretion of government employees, not as a replacement for their intuition, local knowledge and expertise. The optimal use of this tool is when employed as a complement to existing practices.
Given that governments are historically averse to change, predictive analytics will also boost confidence in change and encourage experimentation. Ultimately, data-driven predictions will not only improve government responsiveness and results, but also create conditions more conducive to innovation by reducing risk to government innovators. In tandem with executive leadership, cross-agency collaboration and transparency, data will drive better services, better management and better citizen engagement.