Pavement Roughness Prediction Using Explainable and Supervised Machine Learning Technique for Long-Term Performance

Maintaining and rehabilitating pavement in a timely manner is essential for preserving or improving its condition, with roughness being critical factor. Accurate prediction of road vital component sustainable transportation because it helps planners to develop cost-effective maintenance rehabilitation strategies. Traditional statistical methods can be less effective this purpose due their inherent assumptions, rendering them inaccurate. Therefore, study employed explainable supervised machine learning algorithms predict the International Roughness Index (IRI) asphalt concrete Sri Lankan arterial roads from 2013 2018. Two predictor variables, age cumulative traffic volume, were used study. Five models, namely Random Forest (RF), Decision Tree (DT), XGBoost (XGB), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), utilized compared model. The findings revealed that algorithms’ predictions superior those regression model, coefficient determination (R2) more than 0.75, except SVM. Moreover, RF provided best among five extrapolation global optimization capabilities. Further, SHapley Additive exPlanations (SHAP) analysis showed both explanatory variables had positive impacts on IRI progression, having most significant effect. Providing accurate explanations decision-making processes black box models using SHAP increases trust users domain experts generated by models. Furthermore, demonstrates use AI-based was traditional prediction. Overall, approach, authorities plan avoid costly extensive rehabilitation. promoted extending life reducing frequent reconstruction.