Inspiration
We were inspired by waiting in line at the Brittain dining hall for over 30 minutes while trying to get lunch last week, and when we finally got to the front there was little food left, and we had to wait an additional 10 minutes for fresh food to be brought out.
What it does
The app utilizes a neural network and predicts what times the dining halls are the busiest and how low on food they are. By using user input and predefined data, the neural network trains itself to better predict dining hall conditions over time. We hope to provide students with a reliable way to avoid the busiest times and long waits.
How we built it
We used Android Studio and Java to build the framework of the app and Python to write the machine learning required to build the neural network.
Challenges we ran into
Our largest challenge was communicating between the front-end Java logic and the back-end Python neural network.
Accomplishments that we're proud of
We are very proud of our successful implementation of a neural network based off of over 10,000 distinct data points instead of a basic linear regression model along with efficiently storing user data through the use of CSV files. And while not as technical, our visual design is also something we are quite proud of.
What we learned
We learned how to incorporate a variety of different languages together into a single visual application. We also improved our Android Studio development skills and, for the first time, managed large CSV file to efficiently store user data into the limited memory space of an app.
What's next for GT Dining Detective
In the future, we would like to use a Cloud service to store the data of a large number of users, along with utilizing Google's API to improve the accuracy of our app. We would also like to port the app to IOS in the future.
Log in or sign up for Devpost to join the conversation.