Data Mining Methods for Assessing Self-Regulated Learning

This chapter discusses the recent advancements and potential for advancement in leveraging data mining methods to study SRL behaviors and processes. It reviews how educational data mining (EDM), conducted on fine-grained data from learner interactions, can produce an understanding of SRL and the phenomena which compose it. Within the broad range of data mining methods used in educational domains, five stand out in their use to detect and study SRL: feature engineering, prediction modeling, sequence mining, cluster analysis, and correlation mining. The chapter defines each of these and their use in research projects. To capture the SRL behaviors, researchers require techniques for detecting key aspects of cognition, metacognition, affect, and motivation in the context of the learning task and the environment. Such analyses often rely upon identifying and assessing learners' cognitive skill proficiency, interpreting their action sequences in terms of learning strategies, detecting relevant aspects of their affect and engagement, and evaluating the students' success in accomplishing their current tasks.