Automatic student engagement measurement using machine learning techniques: A literature study of data and methods

Student engagement is positively related to learning outcomes. The student engagement measurement is studied in varied settings, from the traditional classroom to online learning. Artificial intelligence and machine learning advancements have fueled automatic student engagement analysis. The automated student engagement measurement employed several sensor data such as audio, video, and physiological signals in different settings. This paper presents a literature review of automatic student engagement measurement in the classroom and online learning settings, including data collection and annotation techniques, methods, and evaluation metrics. First, a generalized methodology for automatic student engagement analysis is discussed. Then we describe various data collection techniques and annotation methods widely used in the literature and detail the limitations and advantages. The state-of-the-art machine learning methods and the evaluation metrics used to test those methods are reviewed. Additionally, we extend our literature review to the insight into the existing datasets for evaluating the automatic student engagement methods and recent developments in the machine learning methods on open-source datasets. Finally, we present a comprehensive comparison of the methods proposed on various public datasets based on evaluation metrics and engagement types.