Visualizing log

In this article, we present the application of a method for visualizing gameplay patterns observed in log-file data from a geometry game. Using VisCareTrails, a data visualization software system based on the principle of timed word trees, we were able to identify five novel behaviors that informed our understanding of how players were approaching the game. We further utilized these newly identified player behaviors by triangulating them with geometry test scores collected from players outside the game setting. We compared the predictive capacity of these behaviors against five demographic characteristics commonly observed to be associated with educational outcomes: age, gender, ethnicity, mother’s education, and attitude toward video games. Two of the novel behaviors we identified, both reflecting inflexible problem-solving strategies, outperformed all demographic variables except age in terms of predicting change in geometry test scores post-gameplay. We believe that this is sound evidence for the utility of VisCareTrails and the timed-word-tree method for identifying pedagogically relevant player behaviors from semi-structured data associated with educational games