PSI - Issue 70

Ch. Vihas et al. / Procedia Structural Integrity 70 (2025) 461–468

467

Training Testing Validation

0.9973

0.9925

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0.9751

0.9502

0.9502

0.9177

0.9163

0.9093

0.8783

0.5303

0.5002

0.4701

Coefficient of Determination (R2)

XG Boost

Random forest

SVM

ANN

ML Models

Fig.4. Variation of Coefficient of Determination of Training, Testing & Validation for ML models

4. Conclusions

The main aim of this study is to predict the concrete compressive strength containing GGBS as a replacement for cement using various machine learning models and to validate them using the validation dataset. Based on the result, it can be concluded as follows. • The training efficiency of an ML model is directly proportional to validation and test accuracy. Hence, more training accuracy will result in better model performance. • It has been observed that machine learning algorithms employing tree-based models and boosting techniques have yielded relatively good results. • The ANN performed exceptionally well, as it can handle complex data well. • The ML models, such as SVM, performed underwhelmingly due to the number of features in the data. The poor training of the SVM model resulted in lower accuracy of the test and validation sets. • Among the four ML models, the XG Boost obtained a better score due to the gradient boosting algorithm. • These machine learning models can also be used to predict the tensile and flexural strength of concrete, and the mineral admixture can be replaced with other materials. References Alyami, M., Nassar, R. U. D., Khan, M., Hammad, A., Siddika, A., Nawaz, R., Fawad, M., & Gamil, Y. (2024). Estimating compressive strength of concrete containing rice husk ash using interpretable machine learning-based models. Case Studies in Construction Materials, 20, e02901. Baral, U., Singh, R. K., & Kumar, K. S. (2024). Application of machine learning in prediction of strength properties of GGBS based geopolymer concrete. In Proceedings of the International Conference on Computational Innovations and Emerging Trends (ICCIET-2024) (pp. 629 – 638). Atlantis Press. Gogineni, A., Panday, I. K., Kumar, P., & Paswan, R. K. (2024). Predictive modelling of concrete compressive strength incorporating GGBS and alkali using a machine-learning approach. Asian Journal of Civil Engineering, 25, 699 – 709.