FRP-RCB ShearPred: An open-source platform for shear capacity prediction of stirrup-free FRP-reinforced concrete beams integrating design codes and ensemble machine learning

Reliable prediction of the shear capacity of stirrup-free FRP-reinforced concrete beams remains challenging because existing design provisions give inconsistent estimates, while most machine-learning studies are implemented as standalone scripts that are difficult to reproduce and retrain. This paper presents FRP-RCB ShearPred , an open-source Python application integrating five international design code formulations and eight regression algorithms within a single retrainable graphical workflow. Applied to a 728-specimen database, all eight ML models outperformed all five design codes under both held-out testing and leave-one-study-out (LOSO) validation across 49 independent experimental programmes, with Random Forest achieving the best LOSO R 2 = 0.618 (RMSE = 44.3 kN) and recommended as the default deployment model.