Forecasting Critical Food Violations in the City of Minneapolis: An Exploratory Case Study of Open Data

To prevent the spread of foodborne illness and meet the requirements of Federal and State food regulations, the City of Minneapolis department of public health conducts over 8,000 food safety inspections of restaurants and grocery stores annually (Minneapolis Finance Department, 2019). The goal of this case study is to explore whether it is possible to prioritize food inspections to find the most hazardous food safety violations sooner by assessing risk with a predictive model. This project uses open data sets published by the city of Minneapolis and weather data from the U.S. National Centers for Environmental Information to develop logistic regression and random forest models. The City of Chicago’s open data food inspection risk model provides the foundation of the approach followed in this case study. Two-dimensional kernel density estimation provides a mechanism for aggregating crime and 311 complaints over a rolling time window prior to the date of inspection. This study finds evidence that a model containing data sourced exclusively from data available in Minneapolis’s food inspection data could be employed to find high risk food violations sooner.

“The capstone experience provided a great opportunity to partner with local government on a project I was interested in. I had a chance to become fluent in Python data tools, explore advanced modeling techniques, and build a model that may help my local community. It was great to have the chance to dive into the code, make mistakes, learn through research, and test my theories”