What is it about?
Food inspections are how public health departments in U.S. cities avoid outbreaks of foodborne illnesses. These departments often collect and maintain data about the food inspections conducted. They also work with limited resources and their objective is to cite the food establishments that harbor unsafe food conditions with violations. In this work, we see one instance of what happens when the departments use this historical data to train a machine learning model and how the model's outcome can result in unfair outcomes for the citizens.
Featured Image
Photo by Alex Haney on Unsplash
Why is it important?
Unfair food inspection schedules can affect citizens differently based on their location of work and residence. Any optimization model that is trained on biased data can lead to such an unfair outcome. It is important to consider data and more broadly the use of optimization in this dynamic setting.
Perspectives
While working on this empirical analysis, I realized the vast set of problems one needs to consider for scheduling food inspections given the time and resource constraints. I also learned that sometimes a simplified approach to public-facing decision-making needs closely monitored, well-tested solutions that consider multiple stakeholders.
Shubham Singh
Read the Original
This page is a summary of: Fair Decision-Making for Food Inspections, October 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3551624.3555289.
You can read the full text:
Contributors
The following have contributed to this page
