Explainable AI in Education: Techniques and Qualitative Assessment
Open accessThe DOI matches its Crossref record; 44 of the 44 references it cites were found in a registry or an index.
Abstract
Many of the articles on AI in education compare the performance and fairness of different models, but few specifically focus on quantitatively analyzing their explainability. To bridge this gap, we analyzed key evaluation metrics for two machine learning models—ANN and DT—with a focus on their performance and explainability in predicting student outcomes using the OULAD. The methodology involved evaluating the DT, an intrinsically explainable model, against the more complex ANN, which requires post hoc explainability techniques. The results show that, although the feature-based and structured decision-making process of the DT facilitates natural interpretability, it struggles to model complex data relationships, often leading to misclassification. In contrast, the ANN demonstrated higher accuracy and stability but lacked transparency. Crucially, the ANN showed great fidelity in result predictions when it used the LIME and SHAP methods. The results of the experiments verify that the ANN consistently outperformed the DT in prediction accuracy and stability, especially with the LIME method. However, improving the interpretability of ANN models remains a challenge for future research.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
A Survey on Bias and Fairness in Machine Learning
Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2022) · ACM Computing Surveys · doi:10.1145/3457607 · Cited by 5,812 · cited 1×
“Judges make decisions on an offender’s release from jail or confinement based on COMPAS [1].”
A Survey on Bias and Fairness in Machine Learning
Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2019) · arXiv · doi:10.48550/arxiv.1908.09635 · Cited by 5,812 · cited 1×
“Judges make decisions on an offender’s release from jail or confinement based on COMPAS [1].”
Machine Learning Interpretability: A Survey on Methods and Metrics
Carvalho, D. V., Pereira, E. M., & Cardoso, J. S. (2019) · Electronics · doi:10.3390/electronics8080832 · Cited by 1,842 · cited 4×
“In order to improve explainability and make sure that forecasts are transparent, comprehensible, and reliable, the majority of the work in the field of ML explanation research is concentrated on developing new methods and approaches [3].”
InfluentialA Review on Predicting Student's Performance Using Data Mining Techniques
Shahiri, A. M., Husain, W., & Rashid, N. A. (2015) · Procedia Computer Science · doi:10.1016/j.procs.2015.12.157 · Cited by 840 · cited 1×
“For instance, Shahiri et al. [8] noted that neural network models had the best prediction accuracy, whereas DTs had the second highest.”
Notions of explainability and evaluation approaches for explainable artificial intelligence
Vilone, G., & Longo, L. (2021) · Information Fusion · doi:10.1016/j.inffus.2021.05.009 · Cited by 598 · cited 1×
“In order to appropriately assist decision-making processes, these explanations should also include useful and actionable information [20].”
Rokach, L., & Maimon, O. (2006) · Data Mining and Knowledge Discovery Handbook · doi:10.1007/0-387-25465-x_9 · Cited by 594
Open University Learning Analytics dataset
Kuzilek, J., Hlosta, M., & Zdrahal, Z. (2017) · Scientific Data · doi:10.1038/sdata.2017.171 · Cited by 549 · cited 1×
“In the fields of learning analytics and educational data mining, the OULAD [22] is a vast and well-known dataset.”
DARPA's explainable artificial intelligence (XAI) program
Gunning, D. (2019) · Proceedings of the 24th International Conference on Intelligent User Interfaces · doi:10.1145/3301275.3308446 · Cited by 546
Artificial Neural Networks for Neuroscientists: A Primer
Yang, G. R., & Wang, X.-J. (2020) · Neuron · doi:10.1016/j.neuron.2020.09.005 · Cited by 376
Ethical challenges and solutions of generative AI: an interdisciplinary perspective
Al-kfairy, M., Mustafa, D., Kshetri, N., Insiew, M., & Alfandi, O. (2024) · Informatics · doi:10.3390/informatics11030058 · Cited by 328 · cited 1×
“Ad-ditionally, as noted by Al-kfairy et al. [41], in order to use AI systems in education, it is essential that ethical concerns be addressed, such as ensuring accountability and fairness in explainable AI.”
Cited by
Niu, T., Liu, T., Luo, Y. T., Pang, P. C., Huang, S., & Xiang, A. (2025) · Scientific Reports · doi:10.1038/s41598-025-12514-5 · Cited by 87
“She then further analyzed the key evaluation metrics of two machine learning models, focusing on an intrinsically interpretable model and post-hoc interpretability techniques 62 .”
Academic Performance Prediction Using Machine Learning Approaches: A Survey
Pan, J., Zhao, Z., & Han, D. (2025) · IEEE Transactions on Learning Technologies · doi:10.1109/tlt.2025.3554174 · Cited by 59
“The methodology in [102] compared DT, an intrinsically explainable model, with ANNs, which require post hocexplainabilitytechniques.”
Zhang, P., Li, Y., Huo, S., Lin, P., Fang, D., Hu, S., Li, B., Xu, Z., Qiu, X., Li, K., et al. (2025) · Water Research · doi:10.1016/j.watres.2025.124098 · Cited by 23
Systematic Literature Review on Explainable Learning Analytics and Educational Data Mining
Gunasekara, S., & Saarela, M. (2025) · IEEE Access · doi:10.1109/access.2025.3643645 · Cited by 25
“Gunasekara and Saarela [149] highlighted that fidelity is a key metric for evaluating explanations in educational prediction tasks.”
Self-citationGunasekara, S., & Saarela, M. (2025) · Information and Software Technology · doi:10.1016/j.infsof.2025.107933 · Cited by 19
Self-citationUrs, A. L. A. D., & Sudharshan, A. (2025) · MethodsX · doi:10.1016/j.mex.2025.103673 · Cited by 11
Aziz, I., Arshad, S., & Hussan, M. A. (2026) · Asian Journal of Psychiatry · doi:10.1016/j.ajp.2026.104851 · Cited by 6
Linguistic Summarization and Outlier Detection of Blended Learning Data
Phong, P. Đ., Pham, L. T. T., & Thanh, T. X. (2025) · Applied Sciences · doi:10.3390/app15126644 · Cited by 3
Artificial intelligence enabled smart education systems
Sharma, A. K., & Kumar, T. V. V. (2026) · International Journal of Systems Assurance Engineering and Management · doi:10.1007/s13198-026-03196-x · Cited by 3
Gunasekara, S., & Saarela, M. (2025) · International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT) · doi:10.1109/ismsit67332.2025.11268007 · Cited by 6
“…must address the well-documented trade-off between model performance and explainability: highly complex models (e.g., deep neural networks) often deliver better predictive accuracy, but their opaque decision-making can undermine trust, hinder adoption, and limit pedagogical usefulness [43].”
Self-citationBoccuzzi, G., Manganello, F., & Nico, A. (2025) · International Conference on Information Technology for Social Good (GoodIT) · doi:10.1145/3748699.3749766 · Cited by 7
From One-Size-Fits-All to AI-Driven Personalization in Education
Salendab, F. A., Baraguir, R. S., Nuñez, R. A. S., & Cogo, D. A. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-2752-5.ch009 · Cited by 2
How Do 13–16-Year-Olds Understand AI? A Topic Modeling Study of Swedish Students’ Perceptions
Saarela, M., Mohammed, A. T., Velander, J., & Zerega, R. (2025) · Cognitive Models and Artificial Intelligence Conference (AICCONF) · doi:10.1109/aicconf64766.2025.11064290 · Cited by 3
“Explainable AI offers insights into the internal workings of AI models, providing students with a more accurate understanding of AI’s decision-making processes [35], [41].”
Self-citationVecchi, E. (2026) · Education Sciences · doi:10.3390/educsci16010149 · Cited by 3
“For the same reason, single decision trees have been excluded from the current analysis, since other studies performed on educational data have already shown that other methods present a superior performance, even at the cost of explainability (Gunasekara & Saarela, 2025).”
Goel, P., & Weber, R. O. (2026) · International Journal of Artificial Intelligence Tools · doi:10.1142/s0218213026400014 · Cited by 1
Huang, L., & Liu, M. (2026) · European Journal of Education · doi:10.1111/ejed.70529 · Cited by 2
Bolatimi, O. S., Ricky-Okine, C., Bolatimi, F. E., Agbesi, C. C. M., Martey, F., Bolatimi, O. R., Peter, A., & Kwaku, E. O. (2026) · Discover Artificial Intelligence · doi:10.1007/s44163-026-02082-6
“Gunasekara and Saarela, [21] found that post hoc explainability techniques, particularly SHAP and LIME, substantially improved the interpretability of complex neural network models without compromising predictive performance.”
Almazroei, E. E. (2026) · Research Square · doi:10.21203/rs.3.rs-8419224/v1
Explainable AI for Student Assessment and Achievement in the Education System
Tabassum, N., Mian, N. A., Almutlaq, A., Anum, L., Javaid, M. U., & Iqbal, M. J. (2025) · International Conference on AI-Driven STEM Education and Learning Technologies (AISTEMEDU) · doi:10.1109/aistemedu67077.2025.11403886
“Researchers like Gunasekara et al. [14] show that although Decision Trees allow transparency in its natural form, Artificial Neural Networks are more accurate and need explainability tools (SHAP and LIME).”
EXPLAINABLE AI FOR STUDENT PERFORMANCE ANALYSIS IN CODING PLATFORMS
CHARITHA, C., & PRASAD, K. C. (2026) · QP-AIDSE · doi:10.65713/qpv2i108
Li, H., Zang, J., & Li, W. (2026) · International Conference on Informatics Education and Computer Technology Applications (IECA) · doi:10.1145/3802133.3802278
The Development of Vocational Education in the Age of Artificial Intelligence
Zhou, W. (2026) · International Conference on Informatics Education and Computer Technology Applications (IECA) · doi:10.1145/3802133.3802295 · Cited by 1
“…(derived from AI-related employment rate, skill transferability, and long-term career sustainability); (3) identify breakthrough factors via SHAP [3]. where l represents the differentiable convex loss function measuring the difference between the predicted systemic adaptability ̂y i and the…”
InfluentialMilanes, S. L. (2026) · International Conference on Machine Learning and Autonomous Systems (ICMLAS) · doi:10.1109/icmlas67792.2026.11483805
“In a recent article in Applied Sciences, XAI methods for education, such as SHAP and qualitative assessment approaches, are synthesized, reinforcing the appropriateness of SHAP for stakeholder-facing interpretation [20].”
Reynoso, M. M., Bibangco, E. J. P., Villamor, M. M., & Moncera, C. J. A. (2026) · International Symposium on Computational and Business Intelligence (ISCBI) · doi:10.1109/iscbi69404.2026.11495761
“These systems commonly leverage regression-based models and more complex machine learning approaches to balance predictive accuracy with interpretability, an essential consideration in educational decision-making contexts [7], [8].”
EdgeLearnX: Explainable and Privacy-Aware Edge-AI Framework for Adaptive Multimodal Remote Learning
Shutharsun, K., Naveen Kumar, S., Mythili Saran, K., & Lawrence, J. (2025) · International Conference on Sustainable Computing and Smart Systems (ICSCSS) · doi:10.1109/icscss64956.2025.11501006
“In a nutshell, the real-time optimization layer makes sure that every learner gets personalized and continuous learning experience irrespective of network, device and cognitive condition [12].”
Sasmitha, M., sasmoko, Hardjanto, V. L., & Hermanus, D. R. (2026) · International Conference on ICT for Smart Society (ICISS) · doi:10.1109/iciss71466.2026.11715143
“Recent studies emphasize that explainable AI (XAI) enhances human-AI collaboration by improving transparency and supporting more effective decision-making processes, enabling users to better understand and utilize AI outputs [30], [31].”
Ethics in Action: Designing Trustworthy Agentic AI for Equitable Learning
Lancaster, E. A., & Marasinghe, V. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-6471-1.ch007
Evaluating AI in Education: Balancing Fairness, Explainability, and Performance
Gunasekara, S., & Saarela, M. (2026) · International Congress on Human-Computer Interaction, Optimization and Robotic Applications (ICHORA) · doi:10.1109/ichora69329.2026.11537052
“In contrast, the ANN’s post-hoc explanations were less consistent, with LIME achieving a fidelity of 80% and stability of 80%, indicating that its local surrogate explanations moderately approximate the behavior of the underlying ANN but may vary across different runs or perturbations [12].”
InfluentialSelf-citationFrom Post-Hoc Regulation to Runtime Governance: Ethics and Explainability in AI-Driven Education
Saarela, M., & Shenoy, P. (2026) · International Congress on Human-Computer Interaction, Optimization and Robotic Applications (ICHORA) · doi:10.1109/ichora69329.2026.11536988 · Cited by 1
“Scholar-ship in AI in education has consistently highlighted trade-offs between predictive performance and interpretability [28], [29], fairness and accuracy [31], and data-driven personalization and privacy protection [34].”
Self-citationFeature Importance Analysis in Student Performance Prediction Using Random Forest and SHAP
Agherai, O., Sraitih, M., Ahriz, S., Aoula, E.-S., Youssfi, M., & Hassani, A. H. E. (2026) · International Conference on Innovative Research in Applied Science, Engineering and Technology (IRASET) · doi:10.1109/iraset68627.2026.11538532
“Besides, In this study [9], an ANN model was applied on the Open University Learning Analytics Dataset (OULAD) achieved an accuracy of 92.09%, in this study, authors have integrated Explainable AI techniques such as SHAP and LIME to enhance the interpretability of the ANN model.”
InfluentialShannaq, B., Almaqbali, S., & Ali, O. (2026) · ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems (ICETSIS) · doi:10.1109/icetsis68266.2026.11548809
“To determine the most significant course design and engagement factors, we supplement the performance reporting with SHAP-based explanations to translate them into lever-like scenarios to support decision making [11], [12], [13].”
InfluentialGunasekara, S., & Saarela, M. (2026) · International Conference on Information and Education Technology (ICIET) · doi:10.1109/iciet69664.2026.11561591
“These dimensions can be in tension with one another: improving explainability may reduce model flexibility [4], [5], while fairness-enhancing techniques can reduce predictive performance [6].”
Self-citationThapa, D. M., & Shakya, S. (2026) · Procedia Computer Science · doi:10.1016/j.procs.2026.06.205
Explainable machine learning for early classification of middle school student performance
Jihaoui, M. E., Abra, O. E. K., & Mansouri, K. (2026) · Frontiers in Education · doi:10.3389/feduc.2026.1877184 · Cited by 1
Trust, transparency, and learning outcomes in AI-mediated language education platforms
Liu, N., & Ingram, C. (2026) · Interactive Learning Environments · doi:10.1080/10494820.2026.2690105
Méndez, Ó. A., & Flórez, H. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-35635-2_8
Explainable Artificial Intelligence in Education: Concepts, Taxonomies, and Challenges
Kayyali, M. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-9675-0.ch002
Implementing Explainable Artificial Intelligence in STEAM Educational Tools
Ramos, M. Y. G., Herrera, L. A. G., & Aguilar-Ortíz, J. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-9675-0.ch004
Akhtar, N. A., Wahidi, N., Goren, A. Y., & Kobya, M. (2026) · Process Safety and Environmental Protection · doi:10.1016/j.psep.2026.109453
Ethical Leadership Through AI Evaluation Frameworks
Lingam, R. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-7411-6.ch012
An Explainable Machine Learning Framework for K–12 Artificial Intelligence Education
Qu, S., & Wang, H. H. (2026) · ASEE Annual Conference & Exposition · doi:10.18260/1-2--58749
Hidayah, I., Kusumawati, R., & Imamudin, M. (2026) · International Journal of Advances in Data and Information Systems · doi:10.59395/ijadis.v7i2.1666
“…a high-performance classification model, but also provides a transparent interpretation of the factors influencing the level of digital divide among students, thereby increasing confidence in the prediction results and supporting data-based decision-making in the field of education [23], [24].”
InfluentialShehada, W., Ashqar, H. I., & Ewais, A. (2026) · Cluster Computing · doi:10.1007/s10586-026-06521-y
Future direction of artificial intelligence in higher education assessment: a systematic review
Yasfi, A. (2026) · Cogent Education · doi:10.1080/2331186x.2026.2728183
Li, N., & Hong, X. (2026) · International Conference on Educational Digitalization and Computer Science in the Guangdong-Hong Kong-Macao Greater Bay Area · doi:10.1145/3828183.3828270
ENHANCING ASSESSMENT LITERACY: A SYSTEMATIC REVIEW OF AI TOOLS IN EDUCATION
Pongkendek, J. J., Rochintaniawati, D., Nahadi, N., Nugroho, E. P., Suana, W., Suparman, A., Saefullah, A., Dewi, W. S., Uliyandari, M., & Lestari, A. K. (2026) · F1000Research · doi:10.12688/f1000research.188623.1
Towards Sustainable Vocational Education: An Explainable Green AI Ecosystem for Workforce Readiness
Novitasari, E., Rahman, K., Kalua, A. L., Viareco, H., Lestari, E., Amalia, R. R., Nuryati, N., & Wahrini, R. (2026) · F1000Research · doi:10.12688/f1000research.190135.1
Amaral, A. A. D., da Silva, H. M., & Challco, G. C. (2026) · Anais do ... Simpósio Brasileiro de Informática na Educação/Anais do Simpósio Brasileiro de Informática na Educação · doi:10.5753/sbie.2026.27292
Big Data-Driven Web-Based e-Collaboration in Translation Teaching
Yang, X., & Lü, F. (2026) · International Journal of Web-Based Learning and Teaching Technologies · doi:10.4018/ijwltt.426046
Kır, D. B. (2026) · Ana Dili Eğitimi Dergisi · doi:10.16916/aded.1969872
Karim, M., Hameed, M., Khan, N., Fida, A., Nauman, M. A., & Jalal, S. (2026) · VFAST Transactions on Software Engineering · doi:10.21015/vtse.v14i4.2443
Metadata
- Authors
- Gunasekara, S., & Saarela, M.
- Affiliations
- University of Jyväskylä (FI)
- Corresponding author
- Gunasekara, Sachini
- Venue
- Applied Sciences
- Publisher
- MDPI AG
- Published
- 2025-01-25
- Volume
- 15
- Issue
- 3
- Pages
- 1239
- ISSN
- 2076-3417 (ISSN-L)
- OpenAlex
- W4406846419
- Cited by
- 53
- References
- 42
- Citations by year
- 2026: 41 · 2025: 12
- FWCI
- 21.02
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Applied Sciences (publishedVersion, cc-by)
- Jyväskylä University Digital Archive (University of Jyväskylä) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Academy of Finland (356314)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Computer Science Applications › Online Learning and Analytics
- Social Sciences › Education › Artificial Intelligence in Education
- Keywords
- explainable AI, artificial intelligence in education, student outcome prediction, decision tree, artificial neural networks, post-hoc explanation, lime, SHAP, model interpretability
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.