Abstract:
An innovative approach was developed as part of the CRAC project 03ABL, “Improving QA&E by Predicting Student Performance Using Data Analytics.” This approach aims to identify at-risk students in various undergraduate (UG) programmes offered by EdUHK. Specifically, it focuses on identifying students likely to receive a degree classification of Third-Class Honours or below based on their academic performance in the first two years of their programme. The objective is to create an automated system that can provide programme leaders (PLs) with a list of at-risk students shortly after the release of secondyear results each August. This early identification will enable programme leaders to intervene and support these students to enhance their understanding and learning, potentially improving their academic outcomes.
Code:
T0282
Principal Project Supervisors:
Keywords provided by authors:
Start Date:
01 Jul 2024
End Date:
31 Dec 2025
Status:
Ongoing
Result:
In the summer of 2024, we used the eAPR system developed by the OCIO to extract the raw student data by programme. From the prior CRAC project, we adapted code that extracts and merges student records into a unified format and code that uses our statistical model to train and test predictions. In 2024/25, we automated the analysis or data from 20 UG programmes (615 current students total, 32 predicted at-risk) and generated reports of student predictions for the PLs of those programmes. The previous year’s results showed a high rate of successful prediction (93.9% correct across 775 students), and a high correlation of predicted with actual GPAs (r=0.882, MSE=0.0168). Reports and guides were uploaded to a portal, and an informational session was delivered to PLs with explanations for how to access and interpret reports and provide feedback. Response forms returned by PLs indicated general agreement with our findings as well as greater interest in carrying out student- rather than course-based interventions. Our team is coordinating with the OCIO to further enhance the automated system that leverages OCIO APIs that handle data intake and code outputs to generate reports that are secure and accessible via our own platform.
Impact:
This project will produce an easy-to-use automatic platform for predicting at-risk students currently enrolled in UG programmes at EdUHK, thereby enhancing our QA procedure and annual programme reporting cum reflection.
Deliverables:
Books/ Book Chapters/ Journal Articles/ Conference Papers
Dong, C., Yip, J. C., Ling, A. M. H., Kwan, J. L. Y., Yu, P. L. H., Lee, A., Yeung, S. S. S., Leung, P. P. W., Yu, E. K. Y., Cheng, E. C. K., Tsui, K. T., Cheng, M. M. H., Lee, J. C.-K., & Li, W. K. (2025). A data-analytical framework for the early detection of at-risk students in higher education. In Proceedings of the 14th International Conference on Educational and Information Technology (ICEIT 2025). Springer Singapore.
Seminars/ Presentations/ Sharing Sessions
Li, W. K., Ling, M. H. A., Yip, J. C., & Dong, C. (2024, 10 April). Project on improving QA&E by predicting at-risk students using data analytics [Online seminar presentation]. Education University of Hong Kong.
Teaching and Learning Resources/ Materials (including online resources)
A project website with a manual for the automated platform for early identification of at-risk students (https://home.eduhk.hk/~cdong/index.html)
Financial Year:
2023-24
Type:
TDG