Blight Prevention

Helping Cities Prevent Blight

The Challenge

Blight is associated with numerous adverse outcomes for residents, including lower quality of life, financial losses when property values decline, higher crime rates, lower service availability, and decreases in public health. Blight is like a disease: if left untreated, it spreads across the neighborhood, depressing values of properties nearby.

Cities prevent blight by attempting to identify and address small risks before they turn into big problems. But this can be a resource-intensive task for small teams of inspectors who are responsible for numerous geographically dispersed properties. Unfortunately, inspectors waste a lot of precious time on relatively low-risk properties while missing other properties that go into blight.

Existing Approach

Many cities rely on reactive building inspections and code enforcement. Inspectors respond to citizen complaints and then work with property owners to bring their buildings into compliance with regulations. In Cincinnati, this approach leads to inspections in only a quarter of homes that eventually become vacated. Thus, a large fraction of at-risk homes are unknown to the building inspectors who are trying to reduce neighborhood blight.

Our Solution

[Blog Post] [Technical Paper] [Code Repository]

We use machine learning to help inspectors choose which properties to inspect. Machine learning is the ability of a computer to learn patterns in data and to use those patterns to make accurate predictions. It has been used over the past 30 years to solve thousands of problems, from driving cars to providing search results on Google. Machine learning can handle complex data from disparate sources, including past inspection records, property records, 311 calls, emergency dispatches, demographics, and crime. This enables not only more accurate predictions but also a better understanding of what puts properties at risk.


Working with the City of Cincinnati, we supplemented city records (inspection data, fire dispatches, crime data, building permits, and 311 calls) with American Community Survey data, property sales, the Hamilton County’s Geographic Information System, and property taxes. We then simulated history by showing how our system would have done if Cincinnati had used it.

Compared to the baseline, our system correctly flags 12-14% more properties that become vacant with the same number of inspections.

What We’re Doing

  1. Testing the system in Cincinnati. We’re helping the Department of Buildings and Inspections conduct a field trial to ensure the model works as expected. The city will then use the model to help choose which properties to inspect.
  2. Looking for new partners. This is a relatively low-cost, low-risk project for new cities to join, now that we’ve written code for this project and demonstrated that it can work. We hope to scale it to new cities, helping to combat blight while sharing lessons learned from one jurisdiction to the next. 

Are You Our Next Partner?

We’d like to partner with more departments that have the data, staff, resources, and willingness necessary to adapt and implement the model. To do this project, you will need at least three years of property-level data:

  • Past violations
  • Past inspections
  • Building permits
  • Tax records
  • Property sales records
  • Crime data

You will get more from the model if you also provide the following:

  • Emergency-dispatch data
  • 311 records

We wrote a short note on computational requirements here. We posted directions on how to dump databases here. You can find copies of our standard contracts here.

Ready to Contact Us?

If you think you fit, please let us know.