Identifying Player Strategies Through Segmentation: An Interactive Process Visualization Approach

Abstract

Identifying learners’ problem-solving strategies from telemetry data is a critical task for serious games. Traditional methods like sequence mining, text replays, and statistical analysis often necessitate labor-intensive manual iterations to configure data appropriately and typically focus only on predominant trends. To improve our understanding of learner behaviors, this paper introduces a novel interactive visualization system that leverages player journeys-node-edge graphs depicting trends in sequences of player actions. We also present player segmentation, a new approach aimed at revealing and representing strategies that might otherwise be ignored, filtered out, or dismissed as outliers. We evaluated the effectiveness of our system through a mixed-methods study with 12 participants from our target demographic (game analysts). The results show that segmentation significantly reduces the time needed to identify strategies, suggesting that categorizing data based on causal factors can offer analysts more intuitive and insightful explanations.