Comparative Evaluation of Machine Learning Predictions for Concrete Properties

Abstract

The accurate prediction of concrete properties, such as compressive strength, is crucial for ensuring the safety and efficiency of construction projects. In recent years, machine learning (ML) techniques have emerged as powerful tools for modeling complex relationships between input variables and concrete properties. This study presents a comparative evaluation of various machine learning models, including Support Vector Machine (SVM), Random Forests (RF), Gradient Boost Regression (GBR), XG Boost, Adaptive (ADA) Boost, Light GBR to predict key concrete properties. A comprehensive dataset, consisting of 14 parameters of M30 Grade Concrete was used to train and test the models. The performance of each model was assessed based Mean Square Error and coefficient of determination (R²).

Results indicate that the ADA Boost algorithm performed better in predictions giving 97.09% accuracy and with least Mean Squared Error of about 1.485. Therefore, ADA Boost algorithm can be applied to develop predictive model for assessing the performance of self-healing smart concrete. The findings underscore the importance of selecting the appropriate machine learning model based on the specific characteristics of the data and the desired balance between prediction accuracy and computational efficiency. This research provides valuable insights for engineers and researchers aiming to adopt machine learning approaches in the material science and construction fields, paving the way for smarter, data-driven decision-making in concrete design.