Champion Recommendation System of League of Legends

The multiplayer online battle arena (MOBA) game genre is one of the most popular game genres in the world. One of the MOBA games, League of Legends (LoL), is popular with game players, and professional leagues have been created worldwide. LoL has the ban and pick system to ban or pick champions before the game starts. This system affects the outcome of the game, which is even more important in the professional leagues. In this paper, we present a champion recommendation system of LoL. Recommendations are conducted sequentially in the same manner as in the ban and pick system. To address this challenge, we collect professional matches to create the ban and pick dataset and use the data to train two models for both teams. Experimental results show that the neural network classifier performs better than the random forest classifier. In addition, we explain the reason that the performances of the two models are different by separately analyzing the sequences.