Chicago and Allstate Partner to Predict Food Safety Violations

With help from an Allstate Insurance research team, the City of Chicago
has reportedly developed a model to predict which food establishments
might fail inspections among the more than 15,000 restaurants within
the city’s jurisdiction. The research teams analyzed nearly 100,000
sanitation inspection reports to create the prediction model, which
assesses the likelihood that a food establishment will commit a critical
violation.

According to a city report, key factors include (i) whether the
establishment has a previous critical or serious violation, (ii) the
three-day average high temperature, (iii) nearby garbage and sanitation
complaints, (iv) nearby burglaries, and (v) the length of time the
establishment has been operating. The city tested the model in a double-blind
retrodiction of September and October 2014, finding that following
its algorithm would have resulted in 69 percent of violations being found
in the first month compared to the 55 percent that the existing inspection
order had yielded. “That is, an additional 37 establishments would have
been cited for violations the first month, as opposed to being discovered
later, potentially after patrons became ill,” the report states. The city
has begun using the model to prioritize inspections while continuing a
2013 initiative to track complaints of food poisoning on social media. See
Chicago Tribune, July 6, 2015.

Issue 571