Predictive modeling of shear strength in fiber-reinforced cementitious matrix-strengthened RC beams using machine learning

This study investigates the predictive modeling of shear strength in fiber-reinforced cementitious matrix (FRCM)-strengthened reinforced concrete (RC) beams through a comprehensive machine learning approach. A diverse dataset encompassing 14 distinct parameters, including beam dimensions, material properties, and reinforcement ratios, was collected from the literature studies and experimental results. The dataset underwent rigorous statistical analysis, and machine learning models including Support Vector Regression, Random Forest Regression, Gradient Boosting Regression, eXtreme Gradient Boosting Regression, and Multi-layer Perceptron Regression were trained and evaluated. Random Forest Regression emerged as the most accurate model, exhibiting an R² score of 0.9855. In addition, SHAP analysis provided insights into the relative importance of each parameter in predicting shear strength, highlighting the dominance of longitudinal reinforcement ratio and effective beam depth. The study contributes to the field by offering a robust predictive tool for shear strength in FRCM-strengthened RC beams, bridging the gap between conventional methods and advanced machine learning techniques. The future scope involves exploring additional parameters, experimental validation, and optimization strategies to further enhance the applicability and reliability of the models.