Identifying players and predicting actions from RTS game replays


This paper investigates the problem of identifying a Real- Time Strategy game player and predicting what a player will do next based on previous actions in the game. Being able to recognize a player's playing style and their strategies helps us learn the strengths and weaknesses of a specific player, devise counter-strategies that beat the player, and eventually helps us to build better game AI.We use machine learning algorithms in the WEKA toolkit to learn how to identify a StarCraft II player and predict the next move of the player from features extracted from game replays. Our results reveal that using a J48 decision tree correctly identifies a specific player out of forty-one players 75% of the time. This decision tree can identify a player within the first nine minutes of the game. Finally, we can correctly predict what action out of fifty possible actions a player will choose to do next 81% of the time. For a player we have already identified, we can correctly predict the action 89% of the time. This work informs our research on identifying and learning from Real-Time Strategy players and building better game AI.

Document Type

Conference Proceeding


BWAPI, Data Mining, Machine Learning, Real-Time Strategy Game, WEKA

Publication Date


Journal Title

28th International Conference on Computer Applications in Industry and Engineering, CAINE 2015