Data Analytics in Engineering Education: Lessons Learned from a Canadian Engineering School

Abstract

Listen

Translate article

We extended our earlier student data analytics work [1] to further analysis of 10 years of 11,000+ engineering undergraduates’ academic records in the Faculty of Applied Science and Engineering at the University of Toronto to uncover underlying factors impacting students’ academic performance upon graduation. We explored the potential of using supervised and non-supervised machine learning algorithms to select the key features that strongly correlate with the identified students’ academic performance. K-means clustering method was used to study math competency on academic performance and the findings indicate factors other than math can significantly influence student performance. We also applied Association Rule to study life-long learning attributes based on the completion of minors and certificates. From the results, we noticed that technical minors choices correlated strongly with specific core engineering programs and there are also marked differences in terms of gender, legal status and Professional Engineering Year/COOP completion. Various visualization graphical methods including Heatmap, Pie Char, Sanky Diagrams were applied to aid the analysis. The power of data analysis enables a better understanding of our engineering students’ experiences and informs evidence-based decision-making in our school.