(209f) Embedding Reflective Activities in a ChE Data Science and Machine Learning Class

Interactive programming modules, such as Jupyter Notebooks and MATLAB Live Scripts, are widely utilized in teaching data science and machine learning techniques. These tools provide a flexible learning environment due to their interactive nature and ability to integrate live code, explanatory text, equations, and visualizations within a single document. They also facilitate easy sharing and dissemination of educational materials. However, most available Jupyter Notebooks on data science and machine learning are designed as tutorials or static documents without active learner engagement—specifically, learner reflection. Consequently, students frequently skim through the material without deeply reflecting on the content. Without built-in activities to promote reflection, these learning modules often deliver unsatisfactory learning outcomes.

Based on the idea that “We do not learn from experience. We learn from reflection on experience”, as John Dewey put it, we propose to address this unmet need by embedding reflective activities into these learning modules. We view reflection as a critical component of experiential learning and an essential process that solidifies the connection between what a learner experienced and the meaning they derived from that experience. Therefore, it is vital that student reflection takes place before, during, and after an experience. To achieve that, our design of reflection activities follows Borton’s “What? So What? Now What?” model for pre-experience reflection, in-action reflection, and post-experience reflection.

Last year, we shared findings from our initial implementation in the course “CHEN 5970/6970/6976: Big Data Analytics and Machine Learning in the Process Industry.” The initial results clearly demonstrated the effectiveness of enhancing student learning through built-in reflective activities. Encouraged by the findings, we refined our approach based on our experience and student feedback, and we tested these enhancements in the same course during Spring 2025. In this talk, we will present the improvements made through this 2^nd offering and evaluate their effectiveness in enhancing students’ understanding of machine learning techniques applied to the process industry.