Education big data and learning analytics: a bibliometric analysis

Big data in education has grown with the digitization of learning environments, enabling the conversion of large volumes of educational data into insights through learning analytics. The study situates big data's 3Vs (volume, variety/structure, velocity) within education and argues that learning analytics can improve traditional educational systems and digital learning quality and accessibility. It notes Industry 4.0 pressures on higher education to adapt programs and courses and highlights how analytics and technological advancements can reshape learning environments. Purpose: to analyze trends and recommendations in education big data and learning analytics using bibliometric analysis, and to visualize current trends across topics using Scopus data. Research questions: 1) What is the distribution of education big data and learning analytics publications in 2012–2021? 2) What are the most relevant journals and authors in this research area? 3) Which countries are most significant in this area? 4) What are the primary research keywords over the last decade? 5) Which subject areas most involve education big data and learning analytics?

Design: Systematic literature review employing bibliometric analysis and meta-analyses guided by the PRISMA 2020 statement. Data source: Scopus database. Topic scope: education big data and learning analytics. Identification: database search using the keywords "Education Big Data" and "Learning Analytics" yielded 885 records. Screening: 252 records remained after excluding documents from 2022; subject areas deemed irrelevant (Business, Management and Accounting; Medicine; Energy; Physics and Astronomy; Biochemistry, Genetics and Molecular Biology; Economics, Econometrics and Finance; Earth and Planetary Sciences; Multidisciplinary; Agricultural and Biological Sciences; Chemistry; Neuroscience; Pharmacology, Toxicology and Pharmaceutics); and document types (conference papers, conference reviews, reviews, editorials). Eligibility: 250 articles assessed; 2 were excluded because the full articles were not in English. Included: 250 publications (2012–2021) for bibliometric analysis. Tools: VOSviewer used for bibliometric mapping and visualization (co-authorship networks; keyword co-occurrence). Outputs examined: annual publication distribution, journals, authors, countries (including co-authorship link strengths), keywords (all-keyword and author-keyword co-occurrence), and subject area distribution.

- Publication trends (2012–2021): Growth from 1 publication in 2012 to 54 in 2021; a dip from 31 (2017) to 26 (2018), followed by increases from 2019 onwards. - Journals: IEEE Access was the most relevant journal (TP=8; TC=63; publisher IEEE). Lecture Notes in Educational Technology (TP=8; TC=19; Springer Nature) and Educational Technology and Society (TP=3; TC=126) were also prominent. Additional highly cited venues included Computers in Human Behavior (TP=5; TC=171). - Authors: According to Scopus, Ben Williamson (University of Edinburgh, UK) was identified as the most productive in this area (Year of first publication 2007; TP=60; TC=1700; h-index=25). Hiroaki Ogata (Japan) had TP=371; TC=3155; h-index=27. Lynne D. Roberts (Australia) had TP=107; TC=1883; h-index=23. Ryan Shaun Baker (Columbia University, US) had the highest citations (TC=7752) and TP=278 among the top authors. - Countries: United States led with TP=59 (most significant institution: City College of New York), followed by United Kingdom (TP=35; University of Aberdeen) and China (TP=22; Capital University of Economics and Business). Co-authorship: United States had the highest total link strength with 14 links, 59 documents, 1544 total citations; United Kingdom also had 14 links with 35 documents and 752 citations. - Keywords (all-keyword co-occurrence): "Big Data" had the highest co-occurrence (Oc=126; link strength=485), followed by "Learning Analytics" (Oc=89). Other frequent terms included "Machine Learning" (Oc=38), "Data Analytics" (Oc=36), "Education" (Oc=36), "Data Mining" (Oc=30), and "Learning Systems" (Oc=28). Author keywords: "Big Data" (Oc=90; 150 links) and "Learning Analytics" (Oc=88) were most frequent; others included "Machine Learning" (Oc=33), "Higher Education" (Oc=32), "Data Analytics" (Oc=14), and "Educational Data Mining" (Oc=14). - Subject areas: Computer Science accounted for 34.6% (152 documents) and Social Sciences 33% (145 documents); Engineering 13.9%; Mathematics 5.0%; Arts and Humanities 4.1%; with Chemical Engineering having the fewest publications (3 documents).

The decade-long increase in publications indicates growing awareness of the importance of education big data and learning analytics. The prominence of IEEE-published work and highly cited studies underscores the role of data-driven approaches in understanding learner behavior and improving learning strategies. Findings affirm that big data implementations and learning analytics can enhance personalized learning, teaching effectiveness, and decision-making. Country analyses show the United States as the leading contributor with strong international co-authorship networks, suggesting significant global influence on the adoption and development of learning analytics. Keyword co-occurrence patterns confirm that "Big Data" and "Learning Analytics" are foundational, closely linked to machine learning, data analytics, and educational data mining, reflecting the field's interdisciplinary nature. Subject area distributions highlight Computer Science and Social Sciences as central, aligning with the computational and pedagogical dimensions of educational data science. Overall, the results support the hypothesis that leveraging educational big data and learning analytics advances e-learning environments, early risk detection, and data-informed educational practices.

The study bibliometrically analyzed 250 Scopus-indexed publications (2012–2021) on education big data and learning analytics, revealing sustained growth, leading venues (notably IEEE Access), influential authors (e.g., Ben Williamson; Ryan S. Baker), and the United States' leadership with strong co-authorship networks. Computer Science and Social Sciences dominate the subject areas. The work concludes that big data and learning analytics are critical skills and infrastructures for improving e-learning, understanding learner behavior, and advancing learning strategies. Anticipated future directions include expansion of personalized and adaptive learning, predictive analytics for early risk detection, integration with AI and machine learning (e.g., automated evaluation, intelligent tutoring), dashboards and visualization for decision support, and broader cross-sector collaboration to drive data-driven decision-making in education.

The study relied solely on the Scopus database, potentially omitting relevant publications indexed elsewhere (e.g., Springer Link, IEEE Xplore, Web of Science) and thereby limiting generalizability. The keyword strategy used only "Education Big Data" and "Learning Analytics"; more specific or additional terms (e.g., "Big Data Analytics", "Educational Data Mining", "Deep Learning") might have yielded a broader and deeper dataset and more nuanced analyses. Two non-English full-text articles were excluded, which may introduce language bias.