Learning Analytics in Massive Open Online Courses

Educational technology has obtained great importance over the last fifteen years. At present, the umbrella of educational technology incorporates multitudes of engaging online environments and fields. Learning analytics and Massive Open Online Courses (MOOCs) are two of the most relevant emerging topics in this domain. Since they are open to everyone at no cost, MOOCs excel in attracting numerous participants that can reach hundreds and hundreds of thousands. Experts from different disciplines have shown significant interest in MOOCs as the phenomenon has rapidly grown. In fact, MOOCs have been proven to scale education in disparate areas. Their benefits are crystallized in the improvement of educational outcomes, reduction of costs and accessibility expansion. Due to their unusual massiveness, the large datasets of MOOC platforms require advanced tools and methodologies for further examination. The key importance of learning analytics is reflected here. MOOCs offer diverse challenges and practices for learning analytics to tackle. In view of that, this thesis combines both fields in order to investigate further steps in the learning analytics capabilities in MOOCs. The primary research of this dissertation focuses on the integration of learning analytics in MOOCs, and thereafter looks into examining students’ behavior on one side and bridging MOOC issues on the other side. The research was done on the Austrian iMooX xMOOC platform. We followed the prototyping and case studies research methodology to carry out the research questions of this dissertation. The main contributions incorporate designing a general learning analytics framework, learning analytics prototype, records of students’ behavior in nearly every MOOC’s variables (discussion forums, interactions in videos, self-assessment quizzes, login frequency), a cluster of student engagement…

💡 Research Summary

The dissertation investigates the integration of learning analytics (LA) within Massive Open Online Courses (MOOCs) by conducting a comprehensive case study on the Austrian iMooX xMOOC platform. It begins by contextualising the rapid growth of educational technology over the past fifteen years, highlighting learning analytics and MOOCs as two of the most influential emerging domains. MOOCs, by virtue of being free and globally accessible, attract massive numbers of participants—often reaching hundreds of thousands—creating unprecedented volumes of interaction data that demand sophisticated analytical tools.

A thorough literature review establishes the state‑of‑the‑art in both fields. Learning analytics is defined as the systematic measurement, collection, analysis, and reporting of data about learners and their contexts, with the aim of understanding and optimizing learning. MOOCs are characterised by scale, heterogeneity of learners, high dropout rates, and a strong emphasis on self‑directed learning. Existing research shows that LA has been applied to predict course completion, segment learners, and personalise interventions, yet most studies rely on limited datasets or single‑course analyses, leaving a gap for a generalisable framework.

To address this gap, the author proposes a five‑stage LA framework tailored to MOOCs: (1) Data acquisition, (2) Data preprocessing, (3) Analytical modelling, (4) Visualization, and (5) Feedback delivery. Each stage specifies technical requirements (e.g., handling time‑zone normalization, missing‑value imputation) and pedagogical considerations (e.g., aligning metrics with learning objectives).

The empirical component uses raw log files from iMooX, which capture four principal activity streams: discussion‑forum interactions, video‑player events, self‑assessment quiz attempts, and login frequency. After cleaning and transforming the logs into event‑sequence datasets, descriptive statistics reveal stark differences between overall enrollee behaviour and that of completers. For instance, completers post significantly more forum messages and exhibit higher video‑completion rates.

For deeper insight, the study applies unsupervised machine‑learning techniques. K‑means clustering combined with hierarchical validation partitions learners into four engagement clusters: high‑engagement, medium‑engagement, low‑engagement, and dropout. High‑engagement learners log in early and frequently, re‑watch video segments (average re‑watch rate >70 %), and actively contribute to forums. Low‑engagement and dropout clusters show sparse login patterns and a video‑drop‑off rate exceeding 50 %.

Natural language processing (NLP) is employed on forum posts to conduct sentiment analysis and topic modelling. Positive sentiment prevalence correlates with higher course‑completion rates, suggesting that affective community dynamics play a crucial role in learner persistence.

The analytical outputs are integrated into an interactive dashboard prototype. Visualisations include time‑based activity heatmaps, Sankey diagrams of learner pathways per cluster, and sentiment trend charts. The dashboard provides instructors and administrators with real‑time, actionable insights.

Based on the findings, the dissertation proposes targeted interventions. For low‑engagement learners, automated reminders encouraging video re‑watch and personalized forum invitations aim to increase social presence. High‑engagement learners are offered advanced challenges or mentorship roles to sustain community vitality. These recommendations illustrate a closed feedback loop where analytics inform pedagogical actions, which in turn generate new data for continuous improvement.

The contributions of the work are threefold: (1) a generalizable, modular LA framework suitable for diverse MOOC platforms; (2) an extensive empirical analysis linking granular learner behaviours (forum activity, video interaction, quiz attempts, login frequency) to engagement outcomes; and (3) a functional prototype that operationalises analytics into real‑time decision support.

In conclusion, the dissertation demonstrates that sophisticated learning‑analytics pipelines can be successfully embedded within massive online learning environments, turning massive, heterogeneous datasets into meaningful pedagogical intelligence. This advances both the theory of analytics‑driven education and the practice of scaling high‑quality, cost‑effective learning for global audiences.