ika | Devpost

Inspiration

musume.

What it does

ika musume

We're taking actual research data and compiling them using our in-genius interpolation formulas (discussed later) to construct a model that can help predict the amount of plastics in our future oceans. With just one input, the user can peek into future Earth to see how much plastic waste now litters the ocean. We also want to raise awareness to the parts of the world that are struggling to manage their plastic waste by showing a heat map of all plastic waste emitted (by country).

How we built it

Frontend

We used the Vite React.js framework with MapBox to develop a functional map for users to interact with. Along with this, we used TailwindCSS with DaisyUI to customize app components.

Backend

This backend uses Rust with Actix Web framework for HTTP routes that send data back and forth to/from the client. We also used the polars framework to convert CSVs into LazyFrames and DataFrames to get a more organized and viewable data stream.

Data Layer

We used two datasets, the Our World in Data and The World Bank. We calculate an approximation of an emission and use a linear regression formula to estimate total waste.

Below is a diagram showing how we used the datasets to get our approximations:

diagram

Challenges we ran into

Most of our issues is the concept. It took us a long time to figure out how we will calculate a linear regression. It's tricky to work with the datasets because they are very limited. We need to make a few assumptions to get the best approximation we possibly can.

The dataset is also limited because it does not include data about certain countries, like many countries in Africa and some European countries. As a result, there's areas that don't have any heat map color in our predictor GUI.

While MapBox was a success in the end, the setup was riddled with poor documentation and weirdly obscure syntaxing. We pulled through, however - hooray!

Accomplishments that we're proud of

We're proud that we got a working and good-looking application. This was probably the hardest Hackathon we've ever done, so we're glad we were able to do this in 24 hours. HackNJIT has constantly inspired us to make bigger and badder things, and this year was certainly no exception!

ika_arch

What we learned

We learned Javascript's MapBox tool and Rust's polars crate. Most of us haven't worked with big data either, so this was a new learning experience for us. Data Science isn't really our strong suit :P