Building and Environment, 297, 114559
Environment|구충완
Interpretable benchmarking of construction equipment carbon emissions using pre-construction stage information: A machine learning and SHAP-based approach
2026International Journals
Research AreaEnvironment
Professor구충완
AuthorsSeo, S., Ahn, S., Yeom, S., Kim, J., Hong, T., Kang, K., and Koo, C. (Corresponding Author)
Publication Year2026
Journal (Volume, Issue, Pages)Building and Environment, 297, 114559
AbstractAccurate forecasting of carbon emissions from construction equipment at the pre-construction stage is essential for proactive greenhouse gas mitigation in the construction industry. However, conventional assessment approaches have limited capability in estimating equipment-related Scope 3 emissions, as such emissions are highly sensitive to site-specific ground conditions that are difficult to quantify in advance. Existing machine learning-based studies primarily rely on general project characteristics and rarely incorporate geotechnical variability that governs equipment-intensive earthwork operations. To address this gap, this study proposes and validates an interpretable machine learning framework for benchmarking construction equipment carbon emissions by integrating general project characteristics with geotechnical attributes. A refined dataset of 96 completed residential building projects undertaken by major Korean contractors was constructed. Five machine learning algorithms were developed and evaluated using leave-one-out cross-validation. An ensemble model combining decision tree and support vector regression achieved the best performance (mean absolute error: 293.16; symmetric mean absolute percentage error: 16.57 %). SHAP-based analysis identified rippable rock thickness, rock strength, and site area as dominant emission drivers. The proposed framework enables transparent early-stage benchmarking and supports targeted, data-driven carbon reduction strategies.
Construction Equipment Carbon EmissionsInterpretable BenchmarkingMachine LearningSHAP AnalysisGeotechnical CharacteristicsPre-Construction
PrevGaps between observed and perceived indoor environmental quality: An integrated assessment in multiple university library spacesNextA systematic review on the integrations of CFD and artificial intelligence for the future perspectives of built environment: Multi-scale approaches from room to city