Predicting Performance of First Year Engineering Students in Calculus by Using Support Vector Machines

Abstract

Mathematics is one of the most important subjects for engineering education. In School of Engineering, students who enter university without basic mathematical knowledge and skills are categorized as mathematically 'at-risk'. The purpose of this study was to predict 'at risk' students by using Support Vector Machine method. Data of Pamukkale University School of Engineering's 434 incoming students of year 2007 were considered in this study. The result shows that students' university entrance examination mathematics, science and Turkish tests scores and students' high school graduation grade point average are important items to predict students' achievement at university calculus I course. SVM is trained with features of 289 students and tested with features of remaining 145 students. 86% of successful students for calculus I course was predicted as true by SVM.