Clustering Steam User Behavior Data using K-Prototypes Algorithm

Abstract

The use of user telemetry to gather player behavioral data on video games can be very beneficial to game developers with a certain business model. With the help of user telemetry in game development, it can provide access to data on user behavior from installed game clients platform such as Steam. These behavioral data can be used to find out the Steam user behavioral patterns on playtime distributions that can be studied by developers in order to have a deeper understanding of the behaviors of their players. In this study, the data are clustered using the k-prototypes algorithm, a combination of k-means and k-modes algorithm that can be used to cluster mixed attributes. The result shows that the clusters represent the types and preferences of the players.