Flipped learning is an instructional strategy that reverses the traditional classroom model. Instead of using class time for direct instruction (like lectures), students first engage with new material on their own—often through online videos —and then use classroom time for interactive activities. This increasingly prevalent approach allows students to learn asynchronously at their own pace and facilitates deeper engagement and personalized learning during class sessions.

In Singapore Polytechnic, students engage with online lecture videos through a learning management system (LMS) as part of their pre-class preparation. However, the LMS platform provides only basic tracking capabilities, showing whether students have accessed the videos but not offering insights into their detailed viewing behaviors. As such, the institution developed an in-house system called LearningANTS to understand these viewing patterns for lecturers to tailor classroom instruction effectively and address student learning needs before class. Roll & Winne (2015) proposed using learning analytics techniques to study real-time trace to better assess and understand engagement “as it unfolds”.

A mixed methods study was conducted on how LearningANTS provided video clickstream data to enhance classroom engagement. Students (n=840) watched lecture videos enrolled in an engineering mathematics module, with their various interactions—such as play, rewind, fast-forward, and pause—captured through LearningANTS. This system generated visual charts displaying key metrics, including the percentage of videos viewed and detailed playback behaviors. Lecturers then interpreted the data to assess engagement levels and identify learning challenges.

Statistical analysis was conducted on the collected system data and these were analyzed together with assessment performance to show how engagement patterns derived from clickstream data correlated with results. Student surveys helped explain interaction motivations such as rewinding to recall, fast-forwarding to skim and pausing to reflect. Lecturer surveys uncovered how lecturers utilized these data-driven insights to customize their teaching strategies.

The triangulated findings suggest that real-time clickstream analytics can serve as an effective diagnostic system. By monitoring student interactions with lecture videos, lecturers can identify learning difficulties early and tailor their instructional approaches to provide targeted support. This adaptive method not only enhances the learning experience but also facilitates more personalized learning to meet individual student needs. Although this study was carried out on an engineering mathematics module, the approach holds promise for broader applications across various disciplines and larger cohorts in other educational settings.

This presentation will share key findings from the study, explore the potential implications for lecturers, and propose how these insights can inform future deployments of video content management systems. By leveraging video analytics, higher education educators can improve flipped learning, design more effective online asynchronous learning and craft more engaging and tailored personalized learning experiences.

Keywords: Video Clickstream, Learning Analytics, Flipped learning, Student Engagement, Engineering Mathematics.