A Data-Driven Approach to Evaluate the Compressive Strength of Recycled Aggregate Concrete
Document Type
Conference Proceeding
Publication Date
1-1-2023
Abstract
This paper examines the application of machine learning (ML) techniques in the prediction of the compressive strength of recycled aggregate concrete (RAC). The ML models are trained on a comprehensive dataset composed of 981 different RAC test results. Four algorithms of multiple linear regression, lasso regression, random forest, and histogram-based gradient boosting are investigated. The ML training workflow consists of careful feature selection and rigorous hyperparameter tuning based on cross-validation on the training set. The prediction accuracy of ML models was measured by calculating R-squared and root mean squared error metrics. Lastly, sensitivity analysis was performed to measure the impact of input features on the RAC compressive strength. The results indicate that histogram-based gradient boosting model results in the highest accuracy among the studied ML algorithms where water to cement ratio and cement content were found to be the most influential parameters.
Identifier
85179846446 (Scopus)
ISBN
[9780784485163]
Publication Title
ASCE Inspire 2023 Infrastructure Innovation and Adaptation for A Sustainable and Resilient World Selected Papers from ASCE Inspire 2023
First Page
433
Last Page
441
Recommended Citation
Barth, Henry; Banerji, Srishti; Adams, Matthew P.; and Esteghamati, Mohsen Zaker, "A Data-Driven Approach to Evaluate the Compressive Strength of Recycled Aggregate Concrete" (2023). Faculty Publications. 2243.
https://digitalcommons.njit.edu/fac_pubs/2243