Predictive Modelling and Explainability of Metakaolin Concrete Compressive Strength: Benchmarking TabPFN Against Boosted Ensemble and Kernel-Based Algorithms

Authors
  • Oluwafemi Omotayo

    Department of Civil Engineering, Federal University of Technology, Akure

Keywords:
Array, Array, Array, Array, Array
Abstract

Predicting concrete compressive strength accurately, particularly in supplementary cementitious material-based systems, remains a critical challenge despite the advances in modelling techniques in the past decade. This is due to concrete’s highly nonlinear physicochemical interactions. The study therefore evaluates how effectively TabPFN, a transformer-based probabilistic foundation model, compares with support vector regression (SVR), categorical boosting (CatBoost), and extreme gradient boosting (XGBoost), in predicting metakaolin (MK) concrete compressive strength. A dataset of 328 records was curated from 14 literature studies on MK concrete compressive strength, processed, analysed and used in training the selected models. Specific input features in the dataset include oxide compositions of metakaolin; the water-binder ratio; the percentage MK addition; the cement, MK, fine aggregate, coarse aggregate, and water contents, and the curing age, while the target was the compressive strength. Bayesian hyperparameter optimization was carried out on SVR, XGBoost, and CatBoost using Optuna SearchCV, while TabPFN was implemented without hyperparameter tuning. The models’ performance was measured via mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R2), and benchmarked. The results revealed that TabPFN demonstrated the highest predictive accuracy with MAE, MSE, RMSE, R2 values of 2.27, 13.54, 3.68, and 0.98. respectively, outperforming the benchmark models despite its untuned configuration. SHAP analysis revealed that curing age and water content were the most dominant contributors to the compressive strength prediction. The findings demonstrate the robustness of TabPFN for concrete property prediction in sustainable cementitious systems.

References

Abdeen, H. et al. (2026) ‘Interdependence of key molar ratios (SiO2/Al2O3 and Al2O3/Na2O) in metakaolin-based geopolymers: phase composition, microstructure and mechanical insights’. Rochester, NY: Social Science Research Network. Available at: https://doi.org/10.2139/ssrn.6282906.

Abunassar, N. and Alas, M. (2025) ‘Optimization of strength and durability properties of rubberized concrete mixtures containing silica fume using Taguchi method’, Construction and Building Materials, 468, p. 140455. Available at: https://doi.org/10.1016/j.conbuildmat.2025.140455.

Adekitan, O., Oyerinde, A.O. and Jaji, M.B. (2015) ‘Comparative Compressive Strength Assessment of Cement Concretes Blended With Two Locally Produced Natural Pozzolans’, Procs 4th Applied Research Conference in Africa. (ARCA) Conference, 27-29 August 2015. 4th Applied Research Conference in Africa. (ARCA) Conference, Ibadan, Nigeria (In: Mojekwu,J.N., Nani G., Atepor, L., Thwala,W.D., Ogunsumi, L., Awere E., Ocran,S.P., and Bamfo-Agyei, E. (Eds)), pp. 130–145.

Ahmed, H.U. et al. (2022) ‘Statistical Methods for Modeling the Compressive Strength of Geopolymer Mortar’, Materials, 15(5). Available at: https://doi.org/10.3390/ma15051868.

Akeke, G.A. et al. (2023) ‘Experimental investigation and modelling of the mechanical properties of palm oil fuel ash concrete using Scheffe’s method’, Scientific Reports, 13(1), p. 18583. Available at: https://doi.org/10.1038/s41598-023-45987-3.

Akiba, T. et al. (2019) ‘Optuna: A next-generation hyperparameter optimization framework’, Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, pp. 2623–2631.

Akin, O.O. et al. (2020) ‘Prediction of the Compressive Strength of Concrete Admixed with Metakaolin Using Gene Expression Programming’, Advances in Civil Engineering, 2020(1), p. 8883412. Available at: https://doi.org/10.1155/2020/8883412.

Alabi, S.A. and Mahachi, J. (2022) ‘Performance assessment of mechanical and durability properties of cupola slag geopolymer concrete with fly and rice husk ashes’, Nigerian Journal of Technological Development, 19(1), pp. 27–38.

Arum, R.C., Arum, C. and Alabi, S.A. (2022) ‘The highs and lows of incorporating pozzolans into concrete and mortar: A review on strength and durability’, Nigerian Journal of Technology, 41(2), pp. 197–211.

B?rbulescu, A. and Hosen, K. (2025) ‘Cement Industry Pollution and Its Impact on the Environment and Population Health: A Review’, Toxics, 13(7), p. 587. Available at: https://doi.org/10.3390/toxics13070587.

Becerra-Duitama, J.A. and Rojas-Avellanda, D. (2022) ‘Pozzolans: A review’, Engineering and Applied Science Research (EASR), 49(4), pp. 495–504.

Chen, J.J. et al. (2020) ‘Cement Equivalence of Metakaolin for Workability, Cohesiveness, Strength and Sorptivity of Concrete’, Materials, 13(7). Available at: https://doi.org/10.3390/ma13071646.

Cheng, D. et al. (2023) ‘Projecting future carbon emissions from cement production in developing countries’, Nature Communications, 14(1), pp. 1–12. Available at: https://doi.org/10.1038/s41467-023-43660-x.

Cui, L. et al. (2021) ‘Application of extreme gradient boosting based on grey relation analysis for prediction of compressive strength of concrete’, Advances in Civil Engineering, 2021.

Dinakar, P., Sahoo, P.K. and Sriram, G. (2013) ‘Effect of Metakaolin Content on the Properties of High Strength Concrete’, International Journal of Concrete Structures and Materials, 7(3), pp. 215–223.

Ding, J.-T. and Li, Z. (2002) ‘Effects of metakaolin and silica fume on properties of concrete’, ACI Materials Journal, 99, pp. 393–398.

Graybeal, B. and Davis, M. (2008) ‘Cylinder or cube: strength testing of 80 to 200 MPa (11.6 to 29 ksi) ultra-high-performance fiber-reinforced concrete’, ACI Materials Journal, 105(6), p. 603.

Guneyisi, E., Gesoglu, M. and Mermerdas, K. (2008) ‘Improving strength, drying shrinkage, and pore structure of concrete using metakaolin’, Materials and Structures, pp. 1–13. Available at: https://doi.org/10.1617/s11527-007-9296-z.

Haider, I. et al. (2025) ‘Investigating the synergistic effects of Metakaolin and silica fume on the strength and durability of recycled aggregate concrete at elevated temperatures’, Scientific Reports, 15(1), p. 29510. Available at: https://doi.org/10.1038/s41598-025-11494-w.

Hollmann, N. et al. (2022) ‘Tabpfn: A transformer that solves small tabular classification problems in a second’, arXiv preprint arXiv:2207.01848 [Preprint].

Hollmann, N. et al. (2025) ‘Accurate predictions on small data with a tabular foundation model’, Nature, 637(8045), pp. 319–326.

Ikumapayi, C.M., Arum, C. and Alaneme, K.K. (2021) ‘Reactivity and hydration behavior in groundnut shell ash based pozzolanic concrete’, Materials Today: Proceedings, 38, pp. 508–513.

Inaty, F.E. et al. (2025) ‘Mechanical and durability performance of metakaolin and fly ash-based geopolymers compared to cement systems’, Results in Engineering, 27, p. 105788. Available at: https://doi.org/10.1016/j.rineng.2025.105788.

Ismail, M.H., Rusly, N.S.M. and Deraman, R. (2020) ‘Strength and Water Absorption of Concrete Containing Metakaolin and Nylon Fiber’, International Journal of Sustainable Construction Engineering and Technology, 11(1), pp. 230–242.

Johari, M.M. et al. (2011) ‘Influence of supplementary cementitious materials on engineering properties of high strength concrete’, Construction and Building Materials, 25(5), pp. 2639–2648.

Li, S., Yang, J. and Zhang, P. (2020) ‘Water?Cement?Density Ratio Law for the 28?Day Compressive Strength Prediction of Cement?Based Materials’, Advances in Materials Science and Engineering. Edited by S.A. Memon, 2020(1), p. 7302173. Available at: https://doi.org/10.1155/2020/7302173.

Lundberg, S.M. et al. (2020) ‘From local explanations to global understanding with explainable AI for trees’, Nature machine intelligence, 2(1), pp. 56–67.

Lundberg, S.M. and Lee, S.-I. (2017) ‘A unified approach to interpreting model predictions’, Advances in neural information processing systems, 30.

Mermerda?, K. et al. (2012) ‘Strength development of concretes incorporated with metakaolin and different types of calcined kaolins’, Construction and Building Materials, 37, pp. 766–774. Available at: https://doi.org/10.1016/j.conbuildmat.2012.07.077.

Mienye, I. and Jere, N. (2024) ‘A Survey of Decision Trees: Concepts, Algorithms, and Applications’, IEEE Access, PP, pp. 1–1. Available at: https://doi.org/10.1109/ACCESS.2024.3416838.

Miller, S.A. et al. (2021) ‘Achieving net zero greenhouse gas emissions in the cement industry via value chain mitigation strategies’, One Earth, 4(10), pp. 1398–1411. Available at: https://doi.org/10.1016/j.oneear.2021.09.011.

Nadeem, A., Memon, S.A. and Lo, T.Y. (2014) ‘The performance of Fly ash and Metakaolin concrete at elevated temperatures’, Construction and Building Materials, 62, pp. 67–76. Available at: https://doi.org/10.1016/j.conbuildmat.2014.02.073.

Narmatha, M. and Kala, D. (2016) ‘Meta kaolin –The Best Material for Replacement of Cement in Concrete’, IOSR Journal of Mechanical and Civil Engineering, 13, pp. 66–71. Available at: https://doi.org/10.9790/1684-1304016671.

Nguyen, L. (2016) ‘Tutorial on support vector machine’, Applied and Computational Mathematics (ACM) [Preprint].

Pillay, D.L. et al. (2022) ‘Engineering performance of metakaolin based concrete’, Cleaner Engineering and Technology, 6, p. 100383. Available at: https://doi.org/10.1016/j.clet.2021.100383.

Pradhan, S.S. et al. (2024) ‘Effects of rice husk ash on strength and durability performance of slag?based alkali?activated concrete’, Structural Concrete, 25(4), pp. 2839–2854. Available at: https://doi.org/10.1002/suco.202300173.

Pranav, S., Lahoti, M. and Gopalarathnam, M. (2023) ‘Concrete Compressive Strength Prediction Using Boosting Algorithms’, in S.B. Singh et al. (eds) Fiber Reinforced Polymeric Materials and Sustainable Structures. Singapore: Springer Nature, pp. 307–315. Available at: https://doi.org/10.1007/978-981-19-8979-7_26.

Prokhorenkova, L. et al. (2018) ‘CatBoost: unbiased boosting with categorical features’, Advances in neural information processing systems, 31.

Ramezanianpour, A.A. and Jovein, H.B. (2012) ‘Influence of metakaolin as supplementary cementing material on strength and durability of concretes’, Construction and Building materials, 30, pp. 470–479.

Shafiq, N. et al. (2015) ‘Calcined kaolin as cement replacing material and its use in high strength concrete’, Construction and Building Materials, 81, pp. 313–323.

Sharaky, I.A. et al. (2021) ‘Experimental and theoretical study on the compressive strength of the high strength concrete incorporating steel fiber and metakaolin’, Structures. Elsevier, pp. 57–67.

Uddin, M.N. et al. (2023) ‘Interpretable XGBoost–SHAP machine learning technique to predict the compressive strength of environment-friendly rice husk ash concrete’, Innovative Infrastructure Solutions, 8(5), p. 147. Available at: https://doi.org/10.1007/s41062-023-01122-9.

Vapnik, V. (2000) The Nature of Statistical Learning Theory. 2nd edn. New York, NY: Springer (Information Science and Statistics). Available at: https://doi.org/https://doi.org/10.1007/978-1-4757-3264-1.

Verma, M. et al. (2023) ‘Analysis of the properties of recycled aggregates concrete with lime and metakaolin’, Materials Research Express, 10(9), p. 095508. Available at: https://doi.org/10.1088/2053-1591/acf983.

Weng, T.-L., Lin, W.-T. and Cheng, A. (2013) ‘Effect of Metakaolin on Strength and Efflorescence Quantity of Cement-Based Composites’, The Scientific World Journal, 2013(1), p. 606524. Available at: https://doi.org/10.1155/2013/606524.

Zhang, J. et al. (2026) ‘Hybrid Explainable Machine Learning Models with Metaheuristic Optimization for Performance Prediction of Self-Compacting Concrete’, Buildings, 16(1). Available at: https://doi.org/10.3390/buildings16010225.

Zhang, W., Liu, D. and Cao, K. (2024) ‘Prediction of concrete compressive strength using support vector machine regression and non-destructive testing’, Case Studies in Construction Materials, 21, p. e03416. Available at: https://doi.org/10.1016/j.cscm.2024.e03416.

Zhong, C. et al. (2024) ‘An investigation on mechanical properties and durability of metakaolin reinforced modified recycled concrete’, Case Studies in Construction Materials, 20, p. e02978. Available at: https://doi.org/10.1016/j.cscm.2024.e02978.

Cover Image
Downloads
Published
2026-06-19
Section
Articles
License

Copyright (c) 2026 FUTA JOURNAL OF ENGINEERING AND ENGINEERING TECHNOLOGY

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Copyright

With the submission of a manuscript, the corresponding author confirms that the manuscript is not under consideration by another journal. With the acceptance of a manuscript, the Journal reserves the exclusive right of publication and dissemination of the information contained in the article. The veracity of the paper and all the claims therein is solely the opinion of the authors not the journal.

How to Cite

Predictive Modelling and Explainability of Metakaolin Concrete Compressive Strength: Benchmarking TabPFN Against Boosted Ensemble and Kernel-Based Algorithms. (2026). FUTA JOURNAL OF ENGINEERING AND ENGINEERING TECHNOLOGY, 20(2), 11-26. https://doi.org/10.51459/futajeet.2026.20.2.612

Similar Articles

31-40 of 79

You may also start an advanced similarity search for this article.