Inspired by the sports section we decided to predict a UFC matched based on what we have seen so far from the fighters.
What it does
Ever wondered what would happen if your favorite fighters were to clash in a match. Wait no longer, with our up you can find now who would win. Our app uses ML to predict who will win the next match and shows you a ranking based on our ELO points
How we built it
We created a modern web interface with Flutter. This interface is managed our FASTAPI which holds an scikit learn tree classifier model which makes our predictions.
Challenges we ran into
Deciding what to build: after a lot of brainstorming and scrapping ideas due to lack of data we settled to predict UFC
Finding a good data set: many datasets on the web were either incomplete or irrelevant to our purpose
Cleaning and preprocessing the data: even after finding a suitable dataset the work was not done. We had to deal with many Nan values categorical, and irrelevant data.
Flutter: We were new to flutter and had to do many mistakes before getting a nice interface.
Accomplishments that we're proud of
Creating our own ELO system. Using some maths we managed to add a new feature for our model
Building a modern interface with a new framework
Creating a model with over 92%. This is especially impressive since the ones we found on the web have just 70% accuracy