ChoosyChompers

Inspiration

Our dietary restrictions inspired us: no pork, beef, peanuts, eggs -- and one friend who despises the concept of fruit. This tool makes it easier to filter out the foods we can't/won't eat from the menu to find the best food for our tastes.

The best diet is a well-balanced one, and most foods are okay in moderation. With the food recommendation algorithm, people may feel safer venturing a little outside their comfort zones and trying more new foods when out to eat.

What it does

Analyzes food data and segments users into clusters of similar tastes to help recommend foods
Creates standardized menus in person to make food-picking more organized

How we built it

The web app is built with React, TailwindCSS, and TanStack Router. Sign-in with Google is enabled with OAuth through the Google Cloud API. We used Docker to run the backend, which uses TRPC and Prisma.

Clustering and generating example users was done using Sklearn and Numpy, and will hopefully be accessible to the web app through a Flask API by 8am but...

Challenges we ran into

Making sure the website could run properly on everyone's device
Brainstorming ideas that everyone approves of and is comfortable working with
Researching and learning algorithms for data segmentation and how to identify specific individuals with certain clusters in data
Relearning how to use Flask
Managing conflicting codebases on different devices
Making the UI usable and aesthetic across device sizes

Accomplishments that we're proud of

Building a working segmentation model from only ingredients data as the seed for making fake user data and subsequent clustering and identifying which cluster a constructed user would belong to

The beautiful drawer for restaurant recipe creation, adding food to the menu of a restaurant in a very clean, beautiful way.

What we learned

Data segmentation and Sklearn, React + troubleshooting for app running

What's next for ChoosyChompers

We hope to expand this app for use in more restaurants, perhaps be able to scrape existing online menu data, or provide data on cross-contamination risks.