TEACHActive: A Faculty Professional Development Model with Classroom Sensing and Machine Learning

Evrim Baran
Associate Professor
Iowa State University

Educating tomorrow’s STEM workforce for rapidly changing STEM fields requires novel transformative approaches in engineering pedagogy. There is ample evidence of the effectiveness of active learning strategies in improving student learning and engagement in engineering classrooms, yet, the translation to actual classroom practice has been slow. Research suggests that integrating classroom performance data with faculty members’ professional learning has a positive impact on their adoption of evidence-based teaching strategies. Although expert and peer-observations provide personalized formative feedback to the instructor, these approaches are expensive, unscalable, and not sustainable. Faculty need frequent opportunities for observation, feedback, and reflection on the use of their active learning strategies in their classrooms, but there are no validated automated approaches available. The purpose of this research is to design and implement an innovative professional development model, TEACHActive, that leverages classroom analytics to provide automated observation and feedback on the in-class implementation of various active learning strategies in engineering classrooms. The TEACHActive goes beyond traditional one-size-fits-all models by integrating classroom data with continuous, timely, and formative automated feedback while centering instructors as the stewarding agents of pedagogical innovation. TEACHActive includes three main components (a) training on active learning strategies, (b) automated classroom observation, and (c) feedback in the form of classroom analytics from automated observation followed by reflective prompts. The following research questions guide our research: How can we design automated feedback systems with instructors that align pedagogical strategies with classroom analytics? How does participation in TEACHActive change faculty’s facilitation behavior of active learning strategies? How does data-informed automated feedback promote faculty’s reflective practice?TEACHActive is designed and implemented following the design-based implementation research (DBIR) to rapidly enact, test, and revise the model with engineering instructors. The TEACHActive includes a series of hands-on training sessions conducted before the implementation of automated classroom observation and feedback system. TEACHActive uses EduSense, a computer vision-based classroom sensing system that tracks instructor and student behaviors. TEACHActive utilizes EduSense’s customized classifiers, outputs this data on an automated feedback dashboard, displays the behavioral indicators for each session, and provides a progress display that compares the data between different sessions. Instructors use the automated feedback on the session and the progress displays to reflect on their pedagogical practices and take actions accordingly. Results from the first iteration revealed that instructors identified some of the behavioral indicators to be more meaningful than others. Instructors also perceived classroom analytics displayed on TEACHActive dashboard to be promising in facilitating feedback for their future teaching. The results from this project will contribute not only to the understanding of the integrated professional development model in engineering and the foundational knowledge for faculty support in other STEM fields. The results will have a broader impact across different majors, levels (undergraduate and graduate student training), campuses, and disciplines beyond STEM. Finally, this work will bolster the potential for researchers in STEM education and in other fields to make use of a new data collection approach, and any associated data analytics technique.


Dana AlZoubi, Iowa State University, Ames, Iowa; Anasilvia Salazar, Iowa State University, Ames, Iowa