ComPred

Our Website

Inspiration

We were inspired by large trading firms who use this data, and wanted to democratize this such that anyone from home can easily trade commodities

What it does

It takes in data from the USDA, NOAA, and the Bureau of Labor Statistics & the Bureau of Economics Analysis to create this prediction model with a random forest backend as our ML model. Additionally, we used news api and displayed it such that users can learn more about the commodity we ranked.

How we built it

We built the backend data processing and ml specs with python and the frontend with react, typescript and tailwind.css

Challenges we ran into

We had many challenges in formatting and cleaning the data we got from the governmental websites, and we tried various hyper-parameters to tune our model which took quite some time.

Accomplishments that we're proud of

We are proud that we got a plethora of data and tuned a ml model to our specifications for the common person to use and utilize for their commodity trading endeavors.

What we learned

We learned that it is very difficult to find reliable data sources that are easy to scrape and analyze with. However, we were finally able to complete our product and allow for easy viewing and analysis

What's next for ComPred

We would like to add in a sentiment factor as well, but due to api rate limiting factors we were unable to complete this within the timeframe. Additionally, more data sources for more accurate rankings would be helpful, and also more data regarding different commodities like rare earth materials.