A prediction of engineering students performance from core engineering course using classification

Abstract

All Engineering students in Thailand must complete four core engineering courses which consist of mechanic, material, drawing and computer programming. These four core courses are essential basis and fundamental. Prediction of student academic performance helps instructors develop good understanding of how well or poor students perform. Thus, instructors can take proper proactive evaluation to improve student learning. Students can predict themselves to gain higher performance in the future. This paper focuses on developing a predictive model to predict student academic performance in core engineering courses. A total of 6,884 records has been collected from year 2004 to 2010. Five classification models are developed using decision tree, naïve bayes, k-nearest neighbors, support vector machine, and neural network, respectively. The results show that the neural network model generates the best prediction with 89.29% accuracy.