Allstate using big data to help ID food problems in Chicago

Which of the city’s 15,000 restaurants and vendors are most likely to be the site of foodborne illnesses and should be targeted for a closer look? How can the city identify which establishments likely sell untaxed cigarettes? Which trees should be trimmed to minimize damage to power lines when a storm rolls through?

allstateAllstate, the Northbrook-based insurer, is tapping big data to try to answer those questions.

Earlier this year, Chicago began using a predictive model that Allstate’s quantitative research team helped develop to improve restaurant inspections. It combines and mines data the city already collected or were readily available to more quickly identify restaurants that pose a greater risk for foodborne illness and thus help prioritize inspections.

Tom Schenk, Chicago’s chief data officer, said the city soon planned to issue its own announcement about the food-inspection data program and declined to comment further. His Twitter feed, @ChicagoCDO, on May 14 tweeted a link to a report titled “Food Inspection Forecasting: Optimizing Inspections With Analytics.” Allstate’s participation was cited in the report.

Historically, each Chicago food inspector is responsible for nearly 470 restaurants. Among those, more than 15 percent of inspections result in at least one critical violation. So random inspections might not be the best way to prevent foodborne illnesses.

Allstate’s quantitative research staffers working on the city food project included lead analyst Stephen Collins. His connections with the nonprofit Civic Consulting Alliance, which was also credited in the “Food Inspection Forecasting” report, led to the project with the city.

Initially Allstate scientists asked “what is it we want to try to predict if we were trying to build a predictive model?” recalled Smart. “The aim was to identify critical violations much sooner, so what kind of variables or information would” foreshadow risks at food establishments?

In 2013, Chicago also began monitoring Twitter for posts that include the words “food poisoning” by people who identify themselves as Chicagoans. That initiative continues, Schenk recently told WBEZ.