A Dataset for StarCraft AI \& an Example of Armies Clustering

TL;DR

This paper introduces a comprehensive StarCraft dataset capturing full game states and demonstrates how clustering armies by composition can facilitate strategic reasoning and outcome prediction in RTS games.

Contribution

The paper provides a detailed RTS game dataset and showcases a novel clustering approach for armies to enhance strategic analysis.

Findings

Clustering armies by composition can predict battle outcomes.

The dataset enables analysis of tactics and strategies in RTS games.

Mixture of Gaussian models effectively represent army compositions.

Abstract

This paper advocates the exploration of the full state of recorded real-time strategy (RTS) games, by human or robotic players, to discover how to reason about tactics and strategy. We present a dataset of StarCraft games encompassing the most of the games' state (not only player's orders). We explain one of the possible usages of this dataset by clustering armies on their compositions. This reduction of armies compositions to mixtures of Gaussian allow for strategic reasoning at the level of the components. We evaluated this clustering method by predicting the outcomes of battles based on armies compositions' mixtures components