Prediction of compressive strength of brick masonry using machine learning models

This paper explores predicting brick masonry strength using machine learning models, such as Linear Regression (LRM), Polynomial Regression (PRM), Exponential Regression (ERM), Support Vector Regression (SVR), Decision Trees (DTM), Random Forests (RFM), Artificial Neural Network (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS). The dataset comprises 245 data points on brick, mortar and masonry strength, split into training, validation and testing sets. Python and MATLAB were used to implement the models. DTM and RFM achieved highest determination coefficient (R2) values of 0.97 and 0.965. SVR also demonstrated high R2 value of 0.936. ANN, ANFIS and PRM achieved moderate to strong R2 values, while LRM and ERM displayed limitations. Overall, DTM, RFM and SVR emerged as the top performers, exhibiting strong predictive capability and good generalization ability. These models are suggested for practical applications in construction projects for their superior accuracy in estimating the masonry compressive strength.