Visual exploratory data analysis methods to characterize student progress in intelligent learning environments

Abstract

Tracing the progress of individual learners as they interact with computer-based learning environments using exploratory data analysis methods can be very useful in recognizing, understanding, and classifying students' learning behaviors and performance. The detailed activity logs recorded by a learning environment like Betty's Brain can be the basis for developing traces of student behavior, but they may be difficult to interpret without knowledge of the system's inner workings and architecture. Screen captures also provide trace information, but they typically contain distracting details that are not relevant to the process of interest. Visualization and interpretation of the learner's path is much easier in structured problem solving environments, but linking activities to learning behaviors is more complex in systems like Betty's Brain, where students have much more choice in their knowledge construction task. We have developed visualization schemes for Betty's Brain to trace the learner's progress in their knowledge construction tasks. We describe two of the visualization schemes in this paper, and then discuss how they may (1) help classroom teachers track their students' learning progress as they build their causal maps, and (2) inform the development of feedback rules for future versions of Betty's Brain.