From Interaction to Systems Thinking: A Visual Analytics Approach for Exploring Pupils’ Behavior in a Digital Learning Environment
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
We introduce a visual analytics (VA) approach grounded in a systems thinking (ST) framework to explore pupils' problem-solving behaviors and ST reasoning from interaction log data collected in a digital learning environment. Understanding pupils' interaction behaviors in such environments is essential for uncovering exploration patterns, task-solving strategies, and informing the design of interactive learning tools. However, such interaction logs are often large, heterogeneous, and variable in length, making them difficult to analyze, particularly when the goal is to relate observable behaviors to reasoning processes such as ST. To address this challenge, we introduce an approach that transforms interaction sequences into multidimensional time series capturing temporal progression across behavioral dimensions, then apply motif discovery, similarity analysis, and clustering to identify recurring behaviors and strategy-aligned cohorts, supported by multi-level visual representations for cohort-, pattern-, and individual-level analysis. Our contributions include an ST-grounded VA approach, a motif-based analysis pipeline for interaction logs, an interactive system for exploratory analysis, and a case study demonstrating its usefulness for uncovering behavioral patterns and generating hypotheses about ST reasoning, such as Analysis, Synthesis, and Implementation. We implement the approach in BEHAVE (Behavioral Exploration and Hierarchical Analytics in Visual Environments). The system is evaluated through a series of workshops with domain experts, a digital learning environment designer, and a developer using interaction log data from pupils' use of the \emph{Tracing Carbon} interactive learning environment. Results show that BEHAVE helps experts identify distinct strategy profiles, uncover trial-and-error behaviors, reveal previously unseen ST reasoning patterns supporting hypothesis generation, and inform pedagogical strategies for teaching ST skills.