Early Identification of Dropout Students in Massive Open Online Courses
dc.contributor.advisor | Cristea, Alexandra | |
dc.contributor.author | Alamri, Ahmed | |
dc.date.accessioned | 2023-06-05T10:16:31Z | |
dc.date.available | 2023-06-05T10:16:31Z | |
dc.date.issued | 2022-12-23 | |
dc.description.abstract | Learning analytics (LA) provides the ability to understand the patterns of students' behaviour and improve their educational outcomes. Today, the capacity to retain more data has contributed significantly to the rapid growth of the field of LA. For instance, Massive Open Online Course (MOOC) platforms offer free courses for millions of students worldwide. Therefore, students who cannot afford the expense of higher education may benefit significantly from the available knowledge in MOOCs. This opens a door for educators and academic researchers with a fascinating variety of learning behaviour data that could be used to analyse students' activities and improve their outcomes. While MOOCs platforms provide knowledge in a new and unique way, the very high number of dropouts is a significant drawback. Several variables are considered to contribute towards learner attrition or lack of interest, which may lead to disengagement or total dropout. In the past decade, many researchers have sought to explore the reasons behind learner’s attrition in MOOCs. The jury is still out on which factors are the most appropriate predictors; nevertheless, the literature agrees that early prediction is vital to allow for a timely intervention. This thesis aims to investigate the early prevention of dropout phenomenon in MOOCs by analysing the gaps in the current literature, identifying the under-researched areas, and developing continuous predictive models that can be used in real-time to identify students at risk of dropingout out of MOOCs. The current thesis explores a light-weight approach based on as little data as possible – since different MOOCs store different data on their users – and thus strive to create a truly generalisable method. Several features (e.g., registration date, students' jumping activities, and the times spent on every single task) have been proposed to predict at-risk students from an early stage. This goal was successfully achieved using different approaches such as statistical data analysis, machine learning and data visualisation. The second aim of this thesis is to employ motivational theories, mapping online student behaviour onto them, to analyse the drives and triggers promoting student engagement. This thesis further contributes by building an Engage Taxonomy of MOOC engagement tracking parameters, mapped over four engagement theories: Self-Determination Theory (SDT), Drive, Engagement Theory (ET), and Process of Engagement. The present thesis shows for the first-time metrics for measurable engagement in MOOCs, including specific measures for Autonomy, Relatedness and Competence. It also evaluates the parameters based on existing (and expanded) measures of success in MOOCs: Completion rate, Correct Answer ratio and Reply ratio. | |
dc.format.extent | 197 | |
dc.identifier.uri | https://hdl.handle.net/20.500.14154/68274 | |
dc.language.iso | en | |
dc.subject | Learning analytics | |
dc.subject | Visualisation | |
dc.subject | MOOCs | |
dc.subject | Behavioural pattern | |
dc.subject | Machine learning | |
dc.title | Early Identification of Dropout Students in Massive Open Online Courses | |
dc.type | Thesis | |
sdl.degree.department | Computer Science | |
sdl.degree.discipline | Artificial intelligence | |
sdl.degree.grantor | Durham University | |
sdl.degree.name | Doctor of Philosophy |