Front-Loading Cost Estimation in Early-Stage BIM Design

This technical report presents a novel approach to cost estimation in early-stage Building Information Modeling (BIM) design. It addresses the challenges faced when applying artificial intelligence (AI) models for cost estimation, particularly the issues arising from the under-representation of early-stage data in training datasets. The authors propose a t-dependent origin anchoring transformation, which is a technique designed to control extrapolation in unseen input distributions while maintaining estimation accuracy in known distributions. The report details the validation of this method, which shows a significant reduction in extrapolation errors and improved cost estimation accuracy across multiple BIM simulations. The introduction outlines the importance of early-stage cost feedback in decision-making and the limitations of existing estimation methods, which often rely on finalized project data. The report concludes with a discussion on the implications of this research for enhancing decision-making in early BIM design stages, emphasizing the need for reliable estimators that can adapt to evolving project conditions.