UFC Predictor Web App
What does the app do?
Have you ever wanted to know who would win in a UFC fight? With prdct you can. It leverages machine learning powered by TensorFlow to help predict who will most likely win in a matchup between two fighters
How was the ML trained
The Machine Learning model was trained using TensorFlow. It takes in a bunch of data points for each fighter including, wins, losses, age, height, reach, significant strike accuracy, submission and takedown averages and many more. This data is preprocessed and then given to the model which I have tuned and based upon the test data it correctly predicts matches around 73-78% of the time. This is statistically significant compared to just guessing and getting 50% correct.
How was it integrated into the app
Using flask I was able to integrate the backend prediction into the frontend React app. Whenever the user submits a fight that they want to be predicted the frontend makes a post request to the backend to return the winner and based upon the pre-trained model it returns who it predicts to be the winner.
The problem it solves
This app allows for people with no UFC knowledge to see who is likely to win. It lets them easily match up fighters to see who is better and allows an easy entry way into the world of combat sports for newcomers.
Log in or sign up for Devpost to join the conversation.