Work in progress — Prediction of students' academic performance in an introductory engineering course

Abstract

Prediction of student academic performance helps instructors develop a good understanding of how well or how poorly the students will perform, so instructors can take proactive measures to improve student learning. This paper reports our recent ongoing efforts that focus on developing a predictive model to predict students' academic performance in an introductory engineering course titled Engineering Dynamics. A total of 2,151 data points were collected from 239 undergraduate students in three semesters. Four predictive models were developed using multivariate linear regression (MLR), multilayer perceptron (MLP) neural networks, radial basis function (RBF) neural networks, and support vector machines (SVMs), respectively. The results show that in many cases, the support vector machine model generates the overall best predictions: The average prediction accuracy is 89.0%-90.9% and good predictions are 62.3%-69.0%.