Professor Jun Feng
Title: Exploring Data-driven Programming Competency model for CS1: An Empirical Study via Five-channel Sequence
Bio:
Prof. Feng Jun received a Ph.D. from City University of Hong Kong in 2006. She is vice faculty dean in the School of Information Science and Technology at Northwest University. Her research areas include pattern recognition and machine learning, especially in the fields of medical imaging analysis and intelligent education. Recent projects have included medical image analysis with deep learning, and intelligent education based on AI and Brain-Human Interaction. She has reviewed for many journals, including TSP, JIVP, MTAP, JDIM, CJC, JCAD, OPE, and INFPHY. Conferences she has reviewed for include IEEE-VR,MICCAI, SIGCSE, IWCSE, and CompEd. She is a member of IEEE and ACM, and is co-author of more than 200 articles and co-editor of three books.
Abstract:
The CC2020 report underscores the importance of transitioning from knowledge-based to competency-based education. While many studies have developed top-down qualitative models of programming competency, there is a lack of bottom-up data-driven quantitative research. Moreover, most CS1 course designs and assessments do not follow existing competency frameworks, which still mainly focus on knowledge-based approaches. To address these challenges,this study tracks the learning activities of 209 students in a CS1 course, containing 44590 code submissions and 10 formative tests with final exam. Extracted students’ features are categorized into five interwoven channels sequences of student: scores, engagement, code metrics, programming skills, and coding style, determined jointly by knowledge, skills, and dispositions. Furthermore, we use four indicators—entropy, turbulence, proficiency, and resilience—to characterize each student’s sequence properties in each channel. Finally, we visualize the results to create observable patterns, illustrating students’ CS1 programming competency and the evolution of their learning process. This early exploration provides insights and references for constructing empirical programming competency models, using appropriate data granularity, and visualizing observable patterns for future research.