Enhanced clash detection in building information modeling: Leveraging modified extreme gradient boosting for predictive analytics

Abstract

• MXGBoost enhances BIM clash detection across multiple disciplines. • The model demonstrates predictive accuracy with MAE of 0.057 and AUC of 0.972. • Integration improves project coordination and reduces construction errors. • Predictive framework streamlines BIM processes, enhancing cost efficiency. This study introduces an intelligent system that involves the Modified Extreme Gradient Boosting (MXGBoost) algorithm to help improve clash detection in Building Information Modeling (BIM) processes. It will automate the clash detection among the architectural, civil, mechanical, electrical, and plumbing (MEP) disciplines while greatly enhancing the coordination of projects with minimal errors to improve performance in construction projects. The implemented model of the Mean Absolute Error (MAE) of 0.057, a Mean Squared Error (MSE) of 0.0034, an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.972, an R-squared (R²) value of 0.928, and a precision rate of 0.941. The result justifies the model's performance and identifies its potential to streamline predictive analytics in BIM concerning clash detection. This system integrated into the BIM workflow is a severe turn toward cost efficiency and a decrease in most project timelines. This work represents one of the key steps toward automated BIM coordination, heralding the beginning of a new frontier for the industry and providing construction professionals with the much-needed tools to operate such projects effectively.