CDOPS: Complex Dynamics of Online Professional Squads

Interactions within human teams are highly complex, dynamic, and multidimensional. Understanding these dynamics can help improve team efficacy and predict future outcomes. However, machine learning approaches to model these dynamics require a significant amount of data to make meaningful predictions. This paper describes the design and implementation of the Complex Dynamics of Online Professional Squads (CDOPS) dataset, developed for exploring human team dynamics. CDOPS comprises a collection of over 1 million rounds of CounterStrike: Global Offensive (CS:GO) games played in professional tournaments under regulated Counter-Strike servers, providing a foundational dataset for exploring squad dynamics in the context of E-sports. We here provide the context of CS:GO followed by a description of the CDOPS dataset, how it was collected, and how we believe it may be best used.