2012 U.S. Professor of the Year




Professor, OCW advocate, socio-economic diversity champion, one-pony pedagogy assailant, UDL believer, replicating education-research studies supporter.








Predicting Academic Performance Using a Rough Set

Theory-Based Knowledge Discovery Methodology

In an effort to predict student performance in an engineering course, Rough Set Theory (RST) is employed as the core of a knowledge discovery process. Student performance is captured in terms of successful course completion. Therefore, students are classified into two categories: those who pass a course and those who do not. The Rough Set Theory paradigm presented here analyzes each student based on a set of attributes. These attributes are collected through a series of surveys conducted in the first week of the course, allowing for early identification of potential unsuccessful students. Variations of the Rough Set approach are evaluated to determine the one most suited for the particular dataset. The results are promising since the accuracy of student performance prediction presents an Area under the Receiver Operating Characteristic Curve equal to 80%. The benefits anticipated from early identification of weak and/or potentially unsuccessful students will enable educators to engage these students at the onset of the course and enroll them in additional activities to improve their performance.



Related Papers

E. Gil-Herrrera, A. Tsalatsanis, A. Yalcin, A. Kaw, "Predicting Academic Performance Using a Rough Set Theory Based Knowledge Discovery Methodology", International Journal of Engineering Education, pp. 992-1002, Vol. 27 (5), 2011.