Transparent and reliable construction cost prediction using advanced machine learning and explainable AI

Abstract

Accurate construction cost prediction is vital for project management, influencing budgeting, resource allocation, and overall success. This study proposes a comprehensive framework that combines machine learning models, uncertainty quantification through Confidence Intervals, and explainable AI techniques using SHAP (SHapley Additive exPlanations) to enhance transparency and decision-making. Ten machine learning models, including Ridge Regression, Lasso Regression, Elastic Net, K-Nearest Neighbor Regression, and advanced ensemble methods such as XGBoost, CatBoost, and HistGradient Boosting, were evaluated on the RSMeans dataset. Among these, HistGradient Boosting achieved the best performance on the testing dataset. Beyond traditional metrics, Confidence Intervals quantified prediction reliability, and SHAP identified critical cost drivers like “Formwork” and “Tributary Area,” enabling interpretable and robust prediction. This study highlights the potential of machine learning models to revolutionize construction cost estimation by integrating predictive accuracy, uncertainty analysis, and explainability. The proposed framework supports resource efficiency and enables process innovation in cost management. It also contributes to the advancement of sustainable building practices, offering a strong foundation for future research and promoting the adoption of machine learning-based solutions with enhanced transparency and confidence.