TY  - CONF
TI  - Prediction of logistics performance and allocation optimization in construction projects using Catboost, LightGBM, and XGboost based on lean construction principles
C1  - Singapore, Singapore
C3  - Proceedings of the 34th Annual Conference of the International Group for Lean Construction (IGLC 34)
SP  - 227
EP  - 237
PY  - 2026
DO  - 10.24928/2026/0265
AU  - Shidqi, Arif H. N.
AU  - Wibowo, Mochamad A.
AU  - Fatchur, Adian
AU  - Ramdhan, Alfain
AD  - Doctoral Candidate, Department of Civil Engineering, Universitas Diponegoro, Semarang, Indonesia,  haidarsanath@gmail.com, orcid.org/0009-0005-1957-2820 
AD  - Professor, Department of Civil Engineering, Universitas Diponegoro, Semarang, Indonesia,  agung.wibowo@ft.undip.ac.id, orcid.org/0000-0002-5434-9107  
AD  - Professor, Department of Computer Engineering, Universitas Diponegoro, Semarang, Indonesia,  adian@ce.undip.ac.id, orcid.org/0000-0002-1921-9358  
AD  - Bachelor’s, Department of Civil Engineering, Universitas Diponegoro, Semarang, Indonesia,  alfainnaga@gmail.com, orcid.org/0009-0003-7544-9507 
ED  - Hamzeh, Farook
ED  - Poshdar, Mani
ED  - Garcia-Lopez,, Nelly P.
AB  - Supply Chain Management performance and strategic optimization allocation are essential determinants of construction project productivity and operational efficiency. Delays in material distribution, misallocation of workforce, and underused equipment often result in workflow interruptions and waste, which contradict the Lean Construction ideals of continuous flow and waste minimization. While the use of machine learning in construction management has expanded in recent years, research focusing on predictive frameworks for logistics and resource utilization grounded in lean methodology remains scarce. This study introduces a machine learning framework designed to forecast logistics outcomes and resource allocation efficiency in construction projects through the application of CatBoost, LightGBM, and XGBoost algorithms. The dataset comprises variables such as material delivery attributes, workforce capacity, task complexity, and project operating conditions, whereas logistics performance metrics and resource utilization indicators form the prediction targets. Model evaluation employs suitable regression and classification criteria. The experimental results reveal that boosting based techniques can effectively model intricate relationships within construction logistics data. Among the algorithms tested, CatBoost performs most effectively, particularly in managing categorical attributes. The proposed framework enables data driven, proactive decisions to optimize resource deployment and support lean oriented project execution by minimizing idle time, delays, and inefficiencies.
KW  - Construction logistic
KW  - lean construction
KW  - machine learning.
PB  - 
T2  - Proceedings of the 34th Annual Conference of the International Group for Lean Construction (IGLC 34)
DA  - 2026/06/22
CY  - Singapore, Singapore
L1  - http://iglc.net/Papers/Details/2554/pdf
L2  - http://iglc.net/Papers/Details/2554
N1  - Export Date: 19 June 2026
DB  - IGLC.net
DP  - IGLC
LA  - English
ER  -