Development of a Workbench to Address the Educational Data Mining Bottleneck

Abstract

In recent years, machine-learning software packages have made it easier for educational data mining researchers to create real-time detectors of cognitive skill as well as of metacognitive and motivational behavior that can be used to improve student learning. However, there remain challenges to overcome for these methods to become available to the wider educational research and practice communities, including developing the labels that support supervised learning, distilling relevant and appropriate data features, and setting up appropriate cross-validation and configuration algorithms. We discuss the development of an Educational Data Mining (EDM) Workbench designed to address these challenges.