MMORPG player classification using game data mining and K-means

Abstract

Analyzing and understanding the standard of players in virtual environments has been an activity increasingly used by digital game developers and producers. Players not only play, they also consume in-game products, buy game sequences and expansions, make small money transactions, advertise the game to friends, and use the game to socialize with other players. In the case of Massive Multiplayer Online Role-Playing Games (“MMORPG”), the types of players vary, and, by classifying players’ behaviors, it is possible for developers to implement changes which satisfy players in targeted manners which may impact their level of interest and amount of time spent in the game environment. This study suggests that it is possible to identify and classify players via gameplay analysis by using consolidated theories such as Bartle’s archetypes or Marczewski’s types of players, which group players with the k-means algorithm. Below, different studies are presented which group players through different methods: behavioral analysis, questionnaires and game telemetry data analysis. There is also a dedicated section describing the game analytics processes and a session with the results obtained from the analysis of a specific guild from World of Warcraft.