-
-
logo
What it does
Our project predicts ice concentration using physics simulations that rely on the provided HRRR data.
How we built it
The frontend was built with React and Vite, and the backend was done entirely in Python for it's ease in using ML libraries.
Challenges we ran into
The biggest challenge was by far trying to figure out how to generate predictions properly. We ran through a TON of ideas (Check our paper), ranging from ML to more deterministic models to physics simulations. It was a ton of fun but also pretty difficult.
Accomplishments that we're proud of
We're proud of coming up with our very own physics simulation and also a way to tune the hyperparameters; We've never done something like that before and it was very rewarding to see it all come together.
What we learned
A ton about deep learning, machine learning, how to prepare datasets properly, how to work with cluttered and often times confusing datasets, and building out frontends.
What's next for I-SEE
Probably building a bit more on the frontend, and making the backend dynamic (Perhaps it can automatically find new weather datasets and then run it's model on them, then push the results to the frontend?)
