Application of canonical correlation analysis to study the influence of mathematics on engineering programs: a case study

Abstract

Mathematical knowledge is essential to improve the analytical thinking of engineering undergraduates. Exploring more information from existing academic data is an essential aspect of the educational research. The objective of this study is to explore the impact of mathematics performance on different engineering programs. The study was conducted with 626 engineering students from seven different disciplines at the Faculty of Engineering, University of Moratuwa, Sri Lanka. Canonical Correlation Analysis (CCA) was employed to investigate the relationship between mathematics courses and other engineering courses with respect to their disciplines. Results of CCA revealed that the mathematics performance in both semester 1 and 2 influences significantly on the students’ academic performance in Level 2 of the seven engineering disciplines considered. Wilk’s lambda test statistic confirmed that only the first canonical variate pair is significant for all disciplines. The squared canonical correlations of first canonical variate pair indicated that the amount of variance between the mathematics performance and academic performance in Level 2 explained varied among seven disciplines from 42% to 68%. The impact is higher from mathematics in semester 2 than that from semester 1 in all disciplines except for Material Science and Engineering discipline. The explainable variability of student academic performance in Level 2 by the canonical variate of mathematics is varied from 27% to 50% among seven disciplines. Based on preliminary analysis, it can be concluded that the performance in mathematics in Level 1 could indicate the trend towards the student academic performance in all engineering programs.