The ultimate goal of this project is to predict match result of League of Legend by using machine learning.
Through observing KDA of players who have different team role, summoner skills and items of players, map objective achieved by teams and other information all players could know during matches, the model should able to predict match result.
The dataset is from Kaggle.
It contains 180000 ranked NA and EUW region matches (retreived on 1st March). They are all season 7 games. After data cleaning, dataset is shrinked to about 100000 games.
I find using Logistic Regression on players KDA can predict wins of matches with 79% accuracy:
