City of Chicago has limited number of food inspections–approximately 32–to inspect nearly 15,000 food establishments. This project used machine learning to determine which food establishment was most likely to have a critical violations and, therefore, should be inspected first. After conducting a citywide experiment, inspectors found critical violations 25 percent faster.
The entire project is open source and the data is available online. It was one of the first open-source predictive analytics project led by a city which also used the burgoning open data portals. The code can be adapted to other cities or code improvements can be submitted.
Research Materials
- Source code
- Project summary
- Working Paper (forthcoming)
Press & Media
PBS Newshour. Aired 2016-02-11
- The Atlantic “Predictive Policing Comes to Restaurants” January 7, 2016
- Data-Smart City Solutions “Delivering Faster Results with Food Inspection Forecasting” May 19, 2015
- Harvard Business Review “How Cities are Using Analytics to Improve Public Health” September 15, 2014