Data and Script for: Data-Augmented Explainable Machine Learning Framework and User-Centric Tool for Classifying Failure Modes and Predicting Flexural Capacity of FRP-Strengthened RC Beams

This dataset and accompanying scripts provide the complete framework used in this study for predicting the flexural capacity and classifying the failure modes of FRP-strengthened RC beams. The dataset includes the original experimentally compiled database and CTGAN-generated synthetic data used for data augmentation. All input features represent geometric properties, reinforcement details, and FRP strengthening parameters, while the outputs include ultimate flexural capacity and failure mode classification.

The repository also contains fully reproducible Python scripts for model training (including ensemble learning algorithms and Optuna-based hyperparameter optimization), performance evaluation, SHAP-based interpretability analysis, and visualization. In addition, the trained models, preprocessing objects (scalers, encoders, and polynomial transformers), and scripts for generating plots and GUI-based deployment are included to facilitate replication and practical application of the proposed framework.