Strategy Analysis Based on Trajectory Feature Considering Team Movement in Tactical Shooter Games

Abstract

This study defines trajectory-based features in the Tactical Shooter Game Counter-Strike: Global Offensive and evaluates their effectiveness. This game is played in teams of five players. We propose trajectory features for teams. ESTA dataset is used in this study, which includes past professional match data. The defined features are time-considered centroids. Classification is performed using k-means while changing parameters. The changed parameters are the number of clusters and features. Additionally, the classification results are utilized for strategic analysis of professional teams. The results showed that high-performing teams tend to go Mid or A short routes and rarely go A site routes slowly that rely on A long.