Proceedings of the 2013 International Conference on Advanced Information Engineering and Education Science (ICAIEES 2013)

Determining Key (Predictor) Course Modules for Early Identification of Students At-Risk

Daqing Chen, Geoffrey Elliott
Corresponding author
Educational data mining, Student retention, Decision tree induction, Key performance indicator
This paper addresses the problem of early identification of at-risk students, and seeks to determine modules on a given course, referred to as predictor modules, in which a student’s performance is implicitly correlated to the end-of-the-first-year performance of the student. Such predictor modules may therefore be used to predict the likelihood of a student’s year progression. A data mining project has been conducted for this study, and decision tree-based predictive models have been created using various historical records of students’ grades and year progressions. The study reveals that a key predictor module exists, and the pass rate of the key predictor module can be used to predict students’ year progression rate. A set of recommendations is given based on the key predictor module identified from the management point of view in relation to improving student retention. The study also suggests that a students’ performance in a key predictor module can be directly linked to both key performance indicator and key result indicator in course management and student support.
Download article (PDF)