Restaurant Health Inspection Score Prediction

[TheSecMaven]

What problem are we trying to solve?

Health inspectors currently pick restaurants to inspect randomly. Using ML, we can predict restaurants most at risk for failing a health inspection. Once these predictions have been made, Austin Department of Public Health staff can use these predictions to prioritize inspections.

Who will benefit (directly and indirectly) from this project?

Austin Community will be more assured that the food they eat at any restaurant is safe. Restaurant owners will benefit because this system could also be used as a warning/forecasting system. City health inspectors will save time by focusing on only the most at risk restaurants, and can spend more time on other issues if neccesary.

Where can we find any research/data available/articles?

The City of Chicago did this. They were able to predict restaurants most at risk and then get to those restaurants faster. https://github.com/Chicago/food-inspections-evaluation

My team and I used the Chicago model as a basis for developing our own model for the city of Louisville during a hackathon https://github.com/PilgrimShadow/DerbyHacks17

What help is needed at this time?

Need developers who want to see this project through. Once I have a few people on board I will open a repository where we can start the selection of data to use in prediction.

What are the next steps (validation, research, coding, design)?

Research what data Austin has that can be of use. In Chicago, they used Business Licenses,Food Inspections,Crime,Garbage Cart Complaints,Sanitation Complaints,Weather, and Sanitarian Information.

I was unable to implement all of these features but we decided to use 311 call data (available in Austin as well). However these are all excellent predictors in some form or fashion in my opinion. Perhaps Austin has data unique to the city that could be of help?

---

Project management

Checklist for NEW ideas 👶

Hey, you're official! You're now part of the growing civic hacking community in Austin. Here's a few things to get started (a couple you've probably already done).

• Create this idea issue
• Flesh out the who, where, and what questions above
• Start the conversation about this new idea on Slack!
• Respond with at least one update on this issue within the next month

Checklist for ACTIVE projects 🔥

Let's get this project started! When this idea starts taking off, the Projects Core Team will start helping this project's lead(s) out with project management and connecting you to resources you may need. To get there, please complete and check off the following:

• Complete Open Leadership 101
• Create a GitHub repository and Slack channel for work.
• Create a README file in your project repository. This file should help newcomers understand what your project is, why it's important, and kinds of help you're looking for.
• Create issues to describe each task that you plan to do or need help with and how a contributor can get started on that task. You might start and stop a lot, so consider issues as your to-do list.

Checklist for FEATURED Projects 🎉

To have your project FEATURED on Open-Austin.org, complete the following documentation. In past projects, well-documented featured projects have more contributions than other projects.

• In your README, link to the Open Austin Community Participation Guidelines or write your own code of conduct.
• Create file: LICENSE to give your project an open license, allowing for sharing and remixing.
• Create file: CONTRIBUTING.md so others know how they can contribute.

Once all of the above is complete,

• Create an issue on the open-austin.github.io repo with the title `Add [this project] to projects page. An Open Austin leader will review this issue and post your project 🎈

If you get stuck at any point, feel free to reach out to the leadership team on Slack by adding @Leadership to your message. We're here to help you make real changes to our city.

[mscarey]

This will probably remain inactive. But if anybody else picks it up, since it's a machine learning law enforcement project, part of the project ought to be accounting for algorithmic bias.