Development of a Gradient Boosting model for predicting the rutting factor of graphene oxide-modified bitumen at medium and high temperatures

The rutting factor is a key metric that affects bitumen’s resistance to permanent deformation. Experimental determination of this factor, though crucial, is costly and time-intensive, especially for graphene oxide (GO)-modified bitumen. This study explored a Gradient Boosting (CGB) model to predict the rutting factor accurately and efficiently. A dataset of 433 experimental results, including 9 influential parameters, was utilized. The CGB model was refined using cross-validation and its performance evaluated with statistical metrics. It was also benchmarked against five other machine learning models. Sensitivity analysis revealed the initial rutting factor as the most significant input. The CGB model demonstrated superior predictive capability for GO-modified bitumen’s rutting factor. This approach offers a potentially valuable tool for pavement materials engineers, saving time, and resources compared to traditional experimental methods.