A Hybrid Machine Learning Model for Early-Stage Construction Cost Estimation

Le résumé fourni par la source

The accuracy of early-stage construction cost estimation of construction projects is crucial to the successful implementation of building projects and to prevent cost overruns. This paper proposes a prediction framework based on the Whale Optimization Algorithm (WOA) and the Extreme Gradient Boosting (XGBoost) to enhance the accuracy of cost estimation while avoiding the impact of hyperparameter selection on the model's predictive accuracy. First, data pre-processing and analysis are conducted on the cost data of cast-in-place concrete buildings, followed by the selection and dimensionality reduction of feature indicators. Finally, a cost prediction model based on the WOA and XGBoost is established. The established model was compared with the Particle Swarm Optimization (PSO) optimized XGBoost, XGBoost, Random Forest, and Linear Regression. The experimental results show that the prediction error of the WOA-XGBoost model is within$\pm 3 \%$, and the prediction accuracy and generalizability of this hybrid model are better than others.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.