Real and TVAE generated synthetic datasets for Prediction of Tensile Strength and Moisture Susceptibility in Sustainable Asphalt Mixtures

This dataset includes both real and synthetic data related to the indirect tensile strength (ITSD) and moisture susceptibility (TSR) performance of RAS–asphalt mixtures. Experimental data were gathered from several studies covering various mixture types and RAS sources, including relevant input parameters (e.g., aggregate properties, binder content, RAS content, volumetric characteristics) and the corresponding ITSD and TSR outputs. The dataset reflects variability in mixture design and material properties while ensuring consistency in experimental testing protocols. To augment and validate the experimental results, synthetic datasets were created using the Physics-Informed Tabular Variational Autoencoder (PI-TVAE). These datasets were evaluated to ensure their statistical and physical characteristics, including minimum, maximum, mean, standard deviation, and distribution similarity (Kolmogorov–Smirnov test and t-test), closely align with the real data. The dataset is intended to support research in: Development of probabilistic learning models for indirect tensile strength and tensile strength ratio prediction, Uncertainty quantification of RAS–asphalt mixture performance, Automation of asphalt pavement performance evaluation, and Benchmarking of generative modelling approaches in civil engineering materials. All datasets are provided in .xlsx format to ensure reproducibility and facilitate reuse.