Publication: Computation of high-performance concrete compressive strength using standalone and ensembled machine learning techniques
2
0
Issued Date
2021
Resource Type
Language
eng
File Type
application/pdf
ISSN
19961944
Other identifier(s)
2-s2.0-85119719583
Rights Holder(s)
Scopus
Bibliographic Citation
Materials. Vol 14, No.22 (2021)
Suggested Citation
Xu Y., Ahmad W., Ahmad A., Ostrowski K.A., Dudek M., Aslam F., Joyklad P. Computation of high-performance concrete compressive strength using standalone and ensembled machine learning techniques. Materials. Vol 14, No.22 (2021). doi:10.3390/ma14227034 Retrieved from: https://hdl.handle.net/20.500.14740/8139
Author(s)
Abstract
The current trend in modern research revolves around novel techniques that can predict the characteristics of materials without consuming time, effort, and experimental costs. The adapta-tion of machine learning techniques to compute the various properties of materials is gaining more attention. This study aims to use both standalone and ensemble machine learning techniques to forecast the 28-day compressive strength of high-performance concrete. One standalone technique (support vector regression (SVR)) and two ensemble techniques (AdaBoost and random forest) were applied for this purpose. To validate the performance of each technique, coefficient of determination (R2), statistical, and k-fold cross-validation checks were used. Additionally, the contribution of input parameters towards the prediction of results was determined by applying sensitivity analysis. It was proven that all the techniques employed showed improved performance in predicting the out-comes. The random forest model was the most accurate, with an R2 value of 0.93, compared to the support vector regression and AdaBoost models, with R2 values of 0.83 and 0.90, respectively. In addition, statistical and k-fold cross-validation checks validated the random forest model as the best performer based on lower error values. However, the prediction performance of the support vector regression and AdaBoost models was also within an acceptable range. This shows that novel machine learning techniques can be used to predict the mechanical properties of high-performance concrete. © 2021 by the authors. Licensee MDPI, Basel, Switzerland.
Subject(s)
Compressive strength
Decision trees
Forecasting
High performance concrete
Machine learning
Random forests
Regression analysis
Sensitivity analysis
Vectors
High-perfor-mance concrete
High-performance concrete
K fold cross validations
Machine learning techniques
Performance
Random forest modeling
Random forests
Support vector regression models
Support vector regressions
Validation checks
Adaptive boosting
Decision trees
Forecasting
High performance concrete
Machine learning
Random forests
Regression analysis
Sensitivity analysis
Vectors
High-perfor-mance concrete
High-performance concrete
K fold cross validations
Machine learning techniques
Performance
Random forest modeling
Random forests
Support vector regression models
Support vector regressions
Validation checks
Adaptive boosting