Using machine learning to find the best pick in a draft

Hello Everyone!

I just wanted to share with you a project I have been working on during my spare time: https://github.com/vingtfranc/LoLAnalyzer

It is an application that uses machine learning to analyze and evaluate drafts. Basically, I have trained a neural network with hundreds of thousands of high-ranked games in order to predict which team is going to win by simply looking at the bans and the picks. In other words, this neural network is like a very experienced player that can guess what champions are the most adapted to any situation. You can use it during your own drafts to pick better champions or simply as a guide to improve your draft decision-making. You will find more info on the GitHub project page.

I have provided a short documentation and a demo neural network for the patch 7.17 so you can quickly set up and try it out. You also have access to all the tools I have used so you can even build and train your own neural network.

If you encounter any problem using the application, please let me know. Any feedback is of course welcome.

See you in the rift!