Student performance prediction based on course grade correlation

Date

2019-03-15

Authors

Lei, Cheng

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

This research explored the relationship between an earlier-year technical course and one later year technical course, for students who graduated between 2010 and 2015 with the degree of bachelor of engineering. The research only focuses on the courses in the program of Electrical Engineering at the University of Victoria. Three approaches based on the two major factors, coefficient and enrolment, were established to select the course grade predictor including Max(Pearson Coefficient), Max(Enrolment), and Max(Pi) which is a combination of the two factors. The prediction algorithm used is linear regression and the prediction results were evaluated by Mean Absolute Error and prediction precision. The results show that the predictions of most course pairs could not be reliably used for the student performance in one course based on another one. However, the fourth-year courses are specialization-related and have relatively small enrolments in general, some of the course pairs with fourth-year CourseYs and having acceptable MAE and prediction precision could be used as early references and advices for the students to select the specialization direction while they are in their first or second academic year.

Description

Keywords

Academic performance, Pearson Coefficient, Course pair

Citation