Data-driven methods have become central to understanding player behavior in digital games, particularly in the interdisciplinary domain of Human–Computer Interaction (HCI) and game analytics. This paper presents a literature-based review of player modelling techniques that leverage large-scale behavioral telemetry, clustering algorithms, and pattern-mining approaches to analyze and interpret player activity. Rather than conducting new empirical analyses, the review synthesizes insights from existing academic and industrial work to compare how researchers identify play styles, examine engagement patterns, and interpret in-game behavioral traces. Across the surveyed literature, data-driven methods exhibit several key strengths, including scalability, ecological validity, and the ability to uncover latent behavioral structures that are not apparent from small-scale observation alone. At the same time, substantial challenges are evident, such as limited interpretability of clustering results, inconsistent definitions of behavioral categories, threats to construct and external validity, and difficulties in generalising findings across different games, genres, and versions. This paper compares methodological choices across published work, outlines common analytic pipelines, and introduces the TACT framework (Telemetry, Abstraction, Clustering/Correspondence, Translation) as a conceptual lens for structuring player modelling projects.
Barbudhe et al. (Mon,) studied this question.