In this project, we use a neural network to determine the chances of winning a match of League of Legends using only data from champ select and the summoners' previous matches.
Most data was collected through Riot's official API, with overall champion stats (like winrate, ban rate, etc.) collected through champion.gg's API and basic champion stats collected through CommunityDragon.
Using the fast.ai library, we achieved an accuracy of around 73%.