Media Learning Behavior Analysis Research Report Study on Learning Duration, Course Completion Rate, and Grade Distribution

In the modern educational landscape, a quiet revolution has taken place. The classroom is no longer confined to four walls; it lives in the digital space where students log in, watch videos, and submit assignments from anywhere. This shift has generated a massive trail of digital footprints, a phenomenon researchers call educational data mining. It is the practice of looking at the records of what students do online—how long they watch a lecture, how often they log in, and how they interact with peers—to understand how they learn. Alongside this, a field known as learning analysis uses these records to measure and improve the learning environment in real time. The central question driving this work is simple yet profound: can we look at the way a student behaves on a screen and predict whether they will succeed or struggle? Understanding this connection is vital because while online learning offers freedom and convenience, it also comes with a high risk of students dropping out or failing to grasp the material.

A recent study by Yunyi Wang from Fuzhou Melbourne Polytechnic dives deep into this question, focusing on three specific behaviors: how much time a student spends learning, whether they finish the course, and the grades they ultimately receive. The researcher gathered data from online learning platforms, looking at the logs that record every click, every minute of video watched, and every homework assignment submitted. They treated this data not just as numbers, but as a story of student engagement. By organizing this information into clear patterns, the researcher was able to see how these different behaviors connect to one another. They found a strong, positive link between the time a student invests, the likelihood of them finishing the course, and the score they achieve. In other words, the data shows that students who spend more time on the material are far more likely to complete the course and earn higher grades.

The study also examined the tools used to make sense of this data. In the past, researchers relied on basic statistics to describe what happened. Today, however, the field has moved toward more sophisticated methods, including deep learning models. These are computer programs designed to recognize complex patterns in data, much like how the human brain learns from experience. While the study reviews existing research where advanced models have demonstrated high predictive power—such as a model by An Menglei and others that achieved a 95.69% accuracy rate in predicting student performance—the broader trend confirms that these deep learning approaches are becoming the standard for analyzing learning effects. This high level of precision in the field suggests that the digital traces left by students are a reliable indicator of their future success, allowing educators to identify at-risk learners long before they fail.

When the researcher looked closer at the students themselves, they discovered that not everyone learns the same way. By grouping students based on their habits, they identified distinct types of learners. Some were "high-input" students who spent significant time on the platform, completed most tasks, and earned top marks. Others were "low-input" learners who spent very little time, rarely finished assignments, and struggled with the material. There were also "fluctuating" students whose engagement and grades went up and down unpredictably, and a group of "highly efficient" learners who managed to get excellent results with a moderate amount of time. This variety shows that a single approach does not fit all; different students require different kinds of support and attention.

The research also highlighted a gap between how much time students spend online and how well they actually understand the material. While the average student spent over 1,300 minutes watching videos and showed a high rate of course completion, their performance on offline tests was often lower than expected. The data revealed that many students were present on the platform but not fully absorbing the content. They might have watched the videos, but without active engagement or review, the knowledge did not stick. This finding points to a critical issue: simply being online is not enough. The study suggests that the design of the courses themselves plays a huge role. When platforms offer long, uninterrupted lectures, students may lose focus. In contrast, shorter, more interactive segments seem to keep learners engaged for longer periods.

To address these challenges, the study offers clear guidance for the future of online education. For the platforms that host these courses, the recommendation is to build better systems that can track student behavior in real time. These systems should act as an early warning mechanism, alerting teachers when a student stops logging in or falls behind, so that help can be offered immediately. For course designers, the advice is to create more engaging learning environments. This means breaking content into smaller, manageable pieces, adding opportunities for interaction, and providing frequent feedback. Finally, for school administrators, the study suggests that teachers need training on how to use these data tools and how to motivate students who are struggling. By combining better technology with thoughtful teaching design, the goal is to turn the potential of online learning into a reality where more students succeed.

The path forward is clear. The data proves that learning behavior is not random; it follows patterns that can be understood and improved. By paying attention to how students spend their time and by using advanced tools to support them, educators can create online experiences that are not just accessible, but truly effective. The study confirms that when we listen to the story told by the data, we can help every learner find their way to success.