EduBridge Analytica: Bridging Theory and Practice in Educational Data Mining for Action Research

Ahmed MOHAMED; Henry Chukwudi JOHN

doi:10.5281/zenodo.20712838

Authors

Ahmed MOHAMED Department of Computer Engineering, Carnegie Mellon University Africa, Kigali, Rwanda Author
Henry Chukwudi JOHN African Leadership University, Kigali, Rwanda Author

DOI:

https://doi.org/10.5281/zenodo.20712838

Keywords:

Educational Data Mining, Learning Analytics, Action Research, Feature Selection, Educator-Centred Analytics, Dashboard-Based Analytics

Abstract

EDM and LA provide effective techniques to analyse education data, but their practical application for educator-led action research remains limited. In addition, most of the existing approaches presume a technical background and do not provide much support for early stages in the analytical process, such as data preparation and feature selection. This article investigates whether a system-supported EDM-LA workflow can enhance efficiency, consistency, and usability of feature selection for action research.

In this study, a quantitative and qualitative (mixed) methods approach was utilised, where two educators performed feature-selection tasks for three action research questions using both manual methods and a system-supported workflow implemented through the EduBridge Analytica prototype. The quantitative measures were task completion time and feature overlap, whereas qualitative feedback considered usability and cognitive effort. The synthetic post-secondary dataset represented the complexity of real educational data.

The results indicate a dramatic reduction in feature selection time from an average of 342 seconds in manual selection to 45 seconds using system support. There was also high feature overlap between manual feature selection and system-assisted manual feature selection, standing between 85% and 90%. This suggests that the system supported, rather than substituted, educator judgement. The cognitive load of the participants reduced during analysis.

From these results, it is clear that by providing an organisational framework for analytical work through a system-supported process, EDM and LA can become a tool that is not only useful for action research but can also facilitate greater accessibility. This research is a contribution to empirical evidence illustrating that a gap must not only be filled by analytics but also by system design that is cognisant of educational processes regarding data work.