Real-time pavement temperature prediction through ensemble machine learning

Yared Bitew Kebede, Ming-Der Yang, Chien-Wei Huang

Abstract: Highlights•Ensemble machine learning is applied to predict hourly pavement temperature at various depths.•Shapley additive explanation is performed to identify the important parameters.•Robust ensemble ML is selected and compared to commonly used ML and existing models.•Principal component analysis is applied and significantly improves XGBoost's predictive performance.•XGB + PCA effectively predicts real-time pavement temperatures at various depths and facilitates timely preventive maintenance.