by Nattapat Tangniyom and Pruet Boonma

JournalASEAN Journal of Scientific and Technological Reports (AJSTR)
Volume27
Issue5 (September – October 2024)
PublisherThaksin University
ISSN2773-8752 (Online)
AbstractRealm of Valor (RoV) is a famous multiplayer online battle arena (MOBA) game. An average of 25 million games are played daily in Thailand alone. The game also competes in international events, with millions of U.S. dollars in the prize pool. However, the game is very complex and requires a player to have high experience to win. In particular, the hero selection process that each user has to perform at the beginning of each game because the set of selected heroes can affect the game outcome, but there are many heroes to be selected. This paper compares machine learning techniques to predict the winner’s side based on the player’s and opponent’s selection of heroes and the relationship among the selected heroes. Three traditional machine learning techniques, namely, k-Nearest Neighbor, Logistic Regression, and Decision Tree, are compared against ensemble learning with their optimized parameters. The algorithms are evaluated using k-fold cross-validation, and the accuracy of each algorithm is measured. The results show that winning predictions can be improved by considering the relationship among selected heroes. Also, ensemble learning can compete with traditional learning.
SourceA Study on Using Machine Learning to Predict Winner in Multiplayer Online Battle Arena (MOBA) Game
Type keyword