cs.CYFeb 19, 2026

Open Datasets in Learning Analytics: Trends, Challenges, and Best PRACTICE

Authors: Valdemar Švábenský, Brendan Flanagan, Erwin Daniel López Zapata, Atsushi Shimada

Organizations: Faculty of Informatics, Masaryk University, Brno, Czech Republic · Kyoto University, Kyoto, Japan · Kyushu University, Fukuoka, Japan

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Open datasets play a crucial role in three research domains that intersect data science and education: learning analytics, educational data mining, and artificial intelligence in education. Researchers in these domains apply computational methods to analyze data from educational contexts, aiming to better understand and improve teaching and learning. Providing open datasets alongside research papers supports reproducibility, collaboration, and trust in research findings. It also provides individual benefits for authors, such as greater visibility, credibility, and citation potential. Despite these advantages, the availability of open datasets and the associated practices within the learning analytics research communities, especially at their flagship conference venues, remain unclear. We surveyed available datasets published alongside research papers in learning analytics. We manually examined 1,125 papers from three flagship conferences (LAK, EDM, and AIED) over the past five years. We discovered, categorized, and analyzed 172 datasets used in 204 publications. Our study presents the most comprehensive collection and analysis of open educational datasets to date, along with the most detailed categorization. Of the 172 datasets identified, 143 were not captured in any prior survey of open data in learning analytics. We provide insights into the datasets' context, analytical methods, use, and other properties. Based on this survey, we summarize the current gaps in the field. Furthermore, we list practical recommendations, advice, and 8-item guidelines under the acronym PRACTICE with a checklist to help researchers publish their data. Lastly, we share our original dataset: an annotated inventory detailing the discovered datasets and the corresponding publications. We hope these findings will support further adoption of open data practices in learning analytics communities and beyond.

Figures & tables

Conference name (with abbreviation) and URLPublisherICORE rank ( Computing Research and Education, 2023 )First yearAnniversary in 2025
International Learning Analytics and Knowledge Conference ( LAK ) https://www.solaresearch.org/events/lakSociety for Learning Analytics Research (SoLAR) and Association for Computing Machinery (ACM)A201115 th
Educational Data Mining ( EDM ) https://educationaldatamining.org/conferencesThe International Educational Data Mining Society (IEDMS)B200818 th
Artificial Intelligence in Education ( AIED ) https://iaied.org/conferencesThe International AI in Education Society and SpringerA198326 th ( Self, 2016 )
Table 1. The most prominent conferences in each field. Data are bound to the time of writing this paper (submitted February 2025). We note that while LAK and EDM have been held annually since their inception, AIED has been held biannually since 1983, then switching to an annual mode since 2017 ( Self, 2016 ) .
Survey of public datasetsPublication dateDatasets reportedOverlap with datasets across all prior workOverlap with datasets in our survey (see Table 3 )
Romero and Ventura (2020)2020, Jan12— (the first survey)8
Mihaescu and Popescu (2021)2021, Feb4412 (27%), superset of ( Romero and Ventura, 2020 )17
BigData Lab @USTC (2021)2021, Nov183 (17%)14
Lin et al. (2024)2024, Jun109 (90%)9
Xiong et al. (2024)2024, Jul1210 (83%)9
Our paper2025–202617229 (17%)—
Table 2. Comparison of our work to the related surveys reviewed in Section 2.1 , sorted by publication date. We do not list the surveys by Haim et al. ( Haim et al., 2023b ; Haim et al., 2023a ; Haim et al., 2023c ) because they examined a subset of our candidate papers.
Figure 1. PRISMA flow diagram. Generated using the tool by Haddaway et al. (2022) . Diagram showing the progression of selected papers.
Full | Short 2020 2021 2022 2023 2024 Subtotal Total
LAK 5030 4229 3923 4922 6629 246133 379
EDM 3029 2244 2629 1822 2139 117163 280
AIED 4966 4076 4040 5326 4927 231235 466
Subtotal 129125 104149 10592 12070 13695 594531 1125
Total 254 253 197 190 231 1125  
Table 4. Distribution of all the 1,125 published papers (full | short) across the three target conferences throughout the years 2020–2024. In two cases, the text of the proceedings preface listed a slightly incorrect total paper count. Therefore, the numbers below are actual counts of all examined papers. (Each year, LAK published a total of 62–95 papers, EDM 40–66 papers, and AIED 76–116 papers.)
Full | Short 2020 2021 2022 2023 2024 Subtotal Total
LAK 4727 4126 3421 4520 6326 230120 350 (92.3%)
EDM 3028 2244 2629 1721 2138 116160 276 (98.6%)
AIED 4864 3970 3840 5226 4825 225225 450 (96.6%)
Subtotal 125119 102140 9890 11467 13289 571505 1076 (95.6%)
Total 244 (96.1%)242 (95.7%)188 (95.4%)181 (95.3%)221 (95.7%)1076 (95.6%) 
Table 5. Distribution of all the 1,076 candidate (eligible) papers. Same as in Table 4 , the papers are divided into full | short across the three target conferences throughout the years 2020–2024. In addition, the row and column Total include a percentage of the candidate papers compared to the corresponding cell in Table 4 .
Full | Short 2020 2021 2022 2023 2024 Subtotal Total
LAK 40 64 11 71 104 2810 38 (10.9%)
EDM 97 613 67 97 711 3745 82 (29.7%)
AIED 610 77 125 134 155 5331 84 (18.7%)
Subtotal 1917 1924 1913 2912 3220 11886 204 (19.0%)
Total 36 (14.8%)43 (17.8%)32 (17.0%)41 (22.7%)52 (23.5%)204 (19.0%) 
Table 6. Distribution of all the 204 selected papers, that is, those that either used or provided an open dataset. The papers are again divided into full | short across the three target conferences throughout the years 2020–2024. In addition, the row and column Total include a percentage of the candidate papers compared to the corresponding cell in Table 5 .
Figure 2. Distributions of dataset frequency across educational topics and levels of students. Bar chart showing the prevalence of natural sciences topics and K-12 contexts.
Figure 3. Geographical contexts of the dataset collection distributed per continent. This map considers 163 datasets (172 minus the 9 that are not applicable). However, some datasets pertained to multiple countries/continents, so the counts in the map sum to 174. The map was automatically generated in Python using the geopandas module ( den Bossche et al., 2024 ) . World map with highlighted continents, showing the prevalence of dataset origin in North America.
Figure 4. Distribution of number of students into bins. Bar chart showing the various categories of the number of students.
Figure 5. Distribution of number of data points into bins. Bar chart showing the various categories of numbers of data points.
UsageAnalyticCommon metrics (sorted by usage)
137×Supervised MLaccuracy (50×), AUC (45×), F1 (38×), RMSE (19×), recall (17×), precision (13×)
102×Deep learningaccuracy (34×), AUC (30×), F1 (30×), recall (12×), RMSE (10×), MAE (10×)
75×Statisticsinferential tests (30×), correlations (19×)
38×Knowledge tracingAUC (25×), accuracy (13×)
17×Large language modelsF1 (7×), accuracy (6×)
8×Unsupervised MLcustom metrics, such as cluster properties
Table 7. The analytical categories and their most common metrics used with the datasets across the 204 selected papers.
IDDataset name as URL, description, and citationUsage in papers at LAK, EDM, and AIED
1100K Coursera’s Course Reviews – labeled user reviews of online courses ( Adona, 2017 )1× ( Xiao et al., 2022 )
22018 Duolingo Second Language Acquisition Modeling (SLAM) Task ( Settles et al., 2018 ; Settles, 2018 )2× ( Ruan et al., 2021 ; Wu et al., 2020 )
3Academic Performance Evolution for Engineering Students – assessment ( Delahoz-Dominguez et al., 2020 )1× ( Cohausz et al., 2023 )
4Argument Annotated Essays (Nau) – sentences from Portuguese essays ( Nau et al., 2020 )1× ( Ferreira Mello et al., 2022 )
⋆\star 5ASAP-AES (Automated Essay Scoring Prize) – essays and their grades ( Hamner et al., 2012 )7× ( Lu and Cutumisu, 2021 ; Doewes and Pechenizkiy, 2021 ; Doewes et al., 2022 ; Doewes et al., 2023 ; Wang et al., 2021b ; Aomi et al., 2021 ; Yamaura et al., 2023 )
⋆\star 6ASAP-SAS (Short Answer Scoring Prize) – textual answers to questions ( Barbara et al., 2012 )4× ( Funayama et al., 2022 ; Condor et al., 2022 ; Zeng et al., 2023 ; Condor and Pardos, 2024 )
Table 8. All 172 datasets sorted alphabetically. The ⋆\star marks datasets used 3+ times. URL validity last checked on 31 January 2025.
Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

1PPrepare for Open Data Sharing Before Collecting the Data.[Reusable]
□\square Create a research consent form that explains to participants the data anonymization and sharing.
□\square Obtain an IRB approval (if needed). In the application, explain data anonymization and sharing.
2RRefine and Preprocess the Data for Open Use.[Interoperable, Reusable]
□\square Anonymize and clean the dataset, especially removing participants’ sensitive information.
□\square Have an independent human validator to check for issues. If using AI tools, ensure human oversight.
3AAdd Clear Documentation and Metadata.[Findable, Accessible, Interoperable]
Appendix
Table 9. The 8-item PRACTICE guidelines for open data management in LA research. Each item has 2 actionable checklist sub-items. Also, each item includes a note (in gray) regarding which aspect of the FAIR guidelines ( Wilkinson et al., 2016 ; European Commission and Directorate-General for Research and Innovation, 2018 ) it supports.
More in this topic

Educational Data Mining

Educational Data Mining studies computational methods for analyzing learning data to understand learners, improve instruction, and support educational decisions.

See all 72 papers

Explore similar work

CardsList
Feb 23, 2026

Eye-Tracking-while-Reading: A Living Survey of Datasets with Open Library Support

Eye-tracking-while-reading corpora are a valuable resource for many different disciplines and use cases. Use cases range from studying the cognitive processes underlying reading to machine-learning-based applications, such as gaze-based assessments of reading comprehension. The past decades have seen an increase in the number and size of eye-tracking-while-reading datasets as well as increasing diversity with regard to the stimulus languages covered, the linguistic background of the participants, or accompanying psychometric or demographic data. The spread of data across different disciplines and the lack of data sharing standards across the communities lead to many existing datasets that cannot be easily reused due to a lack of interoperability. In this work, we aim at creating more transparency and clarity with regards to existing datasets and their features across different disciplines by i) presenting an extensive overview of existing datasets, ii) simplifying the sharing of newly created datasets by publishing a living overview online, https://t.uzh.ch/1Yh, presenting over 55 features for each dataset, and iii) integrating all publicly available datasets into the Python package pymovements which offers an eye-tracking datasets library. By doing so, we aim to strengthen the FAIR principles in eye-tracking-while-reading research and promote good scientific practices, such as reproducing and replicating studies.

Deborah N. Jakobi, David R. Reich, Paul Prasse +3

Department of Computational Linguistics, University of Zurich. · Department of Computer Science, University of Potsdam. · Department of Informatics, University of Zurich.

Eye Tracking
AskPDF
Jul 30, 2026

DODA: A Database of Datasets for Aesthetics Research

With rapid growth in the fields of empirical and computational aesthetics we have seen a vast increase in large image datasets annotated for aesthetics. As the image databases differ widely in many respects (e.g., different standards for annotation), it can be tedious to find the dataset that fits one's research needs best. The absence of a centralized open-science search system causes additional problems. Currently, researchers typically share dataset links in papers or on diverse platforms like OSF, GitHub or Dropbox. Manually searching for details like image quality and content often requires downloading all datasets. Therefore, we present the Database Of Datasets for Aesthetics (DODA), an intuitive Web application in which researchers can browse all important datasets for aesthetics research. DODA provides general information about these datasets (size, resolution, type of annotation, number of annotators, etc.) and for many of them also precomputed quantitative image properties. We discuss relevant criteria for selecting a suitable dataset with DODA and illustrate the benefits of reusing datasets. Our approach facilitates collaboration across the fields of empirical and computational aesthetics. Keywords: empirical aesthetics, computational aesthetics, machine learning, image annotation, quantitative image properties, Open Science

Lisa Koßmann, Ralf Bartho, Christoph Redies +1

Laboratory of Experimental Psychology, Department of Brain and Cognition, University of Leuven (KU Leuven), Leuven, Belgium · Leuven AI Institute, University of Leuven (KU Leuven), Leuven, Belgium · Experimental Aesthetics Group, Institute for Anatomy I, University Hospital Jena, Germany

Personalized Image Aesthetics AssessmentDataset Discovery
AskPDF
Sep 17, 2026

Edustories: A Collection of Real-world Case Studies from Classroom Practices

Despite the widely recognized potential of AI in education, most prior work has focused on individualized student assistance. In contrast, the majority of educational practice worldwide still takes place in collective classroom settings. To enable researchers to study AI assistance in collective teaching, we introduce Edustories, a dataset of 1,492 teacher-written case studies describing real elementary and high-school classroom situations involving challenging student behavior, pedagogical interventions, and their outcomes. Among many other applications, Edustories enables evaluating LLMs' ability to predict the success of teacher interventions, crucial for providing practicing teachers with useful feedback. Comparing the latest models from four language-model families against expert assessments, we find that current models fall short of human expertise in predicting classroom outcomes; the strongest models reach 58% accuracy compared to 64% of human experts. This gap highlights both the limitations and the emerging potential of AI as assistants for practicing teachers.

Michal Štefánik, Jan Nehyba, Jirina Karasova +4

Faculty of Education, Masaryk University, Czechia · R&D Center for Large Language Models, National Institute of Informatics, Japan

AI in EducationEducational Data MiningGenerative AI EvaluationEducation
AskPDF