Predicting performance of students in a flipped classroom using machine learning: Towards automated data-driven formative feedback

Abstract

Learning analytics (LA) is a relatively new research discipline that uses data to try to improve learning, optimizing the learning process and develop the environment in which learning occurs. One of the objectives of LA is to monitor students’ activities and early predict performance to improve retention, offer personalized feedback and facilitate the provision of support to the students. Flipped classroom is one of the pedagogical methods that find strength in the combination of physical and digital environments - i.e. blended learning environments. Flipped classroom often make use of learning management systems in which video-recorded lectures and digital material is made available, which thus generates data about students’ interactions with these materials. In this paper, we report on a study conducted with focus on a flipped learning course in research methodology. Based on data regarding how students interact with course material (video recorded lectures and reading material), how they interact with teachers and other peers in discussion forums, and how they perform on a digital assessment (digital quiz), we apply machine learning methods (i.e. Neural Networks, Naïve Bayes, Random Forest, kNN, and Logistic regression) in order to predict students’ overall performance on the course. The final predictive model that we present in this paper could with fairly high accuracy predict low- and high achievers in the course based on activity and early assessment data. Using this approach, we are given opportunities to develop learning management systems that provide automatic data-driven formative feedback that can help students to self-regulate as well as inform teachers where and how to intervene and scaffold students. Copyright © 2019 by the International Institute ofInformatics and Systemics.