Multi-player online games have a matchmaking system, where two or more players can meet each other in a virtual room [19][20]. Players do not gain competitive experience when the matchmaking system uses a single parameter because the system cannot recognize players properly. Previous studies carried out several matchmaking methods, namely skill-based, behavior-based, role-based, and latency-based. However, few studies have discussed the computation of multi-parameter aggregation. We propose a matchmaking system that combines three essential things that become competitive matchmaking parameters: skill rating, behavior cluster, and location cluster. The Elo rating system will use to measure player ability. It will combine with player behavior clustering and player location clustering. It combines a statistical approach to managing distributions and unsupervised machine learning to identify players. Grouping players with machine learning methods is a new thing that has been researched in recent years [6][8]. The application of the algorithm with these parameters is carried out on the trivia education game to make it more competitive
Copyrights © 2022