Enhancing Machine Learning Skills Among Undergraduates Through Project -Based Learning: A Mixed Method Approach

Abstract

Preparing engineering graduates with practical machine learning skills is crucial in the current era of artificial intelligence. To enhance practical machine learning skills among electronics engineering students, Project -Based Learning (PBL) pedagogy was implemented. To evaluate the effectiveness of the PBL pedagogy, the proposed study investigated the outcomes and experiences of the stakeholders. The study focused on two research questions “What is the impact of transitioning from traditional approach to PBL approach on Machine Learning skills among undergraduate engineering students?” and “What are the perceived benefits and challenges of implementing Project -Based Learning (PBL) in a machine learning course, as experienced by both undergraduate engineering students and faculty?” In the study participants were purposely sampled and twelve students participated in the Focus Group Discussion (FGD) and three faculty participated in the semi -structured interview. The data from FGD and semi -structured interview was used to arrive at the perceived benefits and challenges faced by the faculty and students by implementing Project -Based Learning (PBL) in a machine learning course through thematic analysis of the data. Further, to perform impact analysis, students’ scores and attendance were considered from control group and experimental group to determine the impact of PBL pedagogy. Student data from two academic years was considered to perform impact analysis. The students in the academic years without PBL as pedagogy is considered as ‘control group’ and students with PBL as pedagogy is considered as ‘experimental group’. The proposed study is an empirical and quasi-experimental study with mixed-methods design aiming at measuring the impact of PBL initiative on students and faculty. For the quantitative analysis, the data from three academic years is considered while for the qualitative analysis, data was collected through semi structured interviews from three faculty members, common across the three course deliveries were considered while six students from two academic years of PBL implementation were purposively sampled to participate in the FGD. Quantitative analysis shows that the transition from traditional mode of delivery to PBL mode of delivery helped students to improve their machine learning skills and the impact was evident through changes in the mark’s distribution range (score band range). The initiative helped the students to translate the machine learning skills into publications. The initiative resulted in 70 Scopus indexed publications and one Q1 journal publication. Qualitative analysis of data from FGDs and interviews revealed the benefits and challenges experienced by students and teachers. The analysis highlighted the importance of processes, resources, scaffolding models, and the teacher's transition from lecturer to mentor within the PBL environment. Notably, the study found that course outcomes extended beyond the immediate course, impacting students' subsequent projects and career interests.