Publications / 2024 Proceedings of the 41st ISARC, Lille, France

A Graph Neural Network Approach to Conceptual Cost Estimation

Hao Liu, Jack C.P. Cheng, Chimay J. Anumba
Pages 815-821 (2024 Proceedings of the 41st ISARC, Lille, France, ISBN 978-0-6458322-1-1, ISSN 2413-5844)
Abstract:

Conceptual cost estimation plays an important role in construction projects since it is the basis for stakeholders to produce financial plans (e.g., establishing project budgets). The current practice, heavily dependent on cost engineers' subjective judgment and manual work, tends to be error-prone and labor-intensive. In response, this paper introduces a Graph Neural Network (GNN) approach to accurate and efficient conceptual cost estimation. Firstly, cost factors impacting construction costs, as well as their relationships, are identified based on literature review to form a graph representation. Afterwards, a GNN model is deployed to predict the construction cost. A real-world dataset from school projects is used for validation. The results show that the proposed approach achieved high accuracy, demonstrating the potential of graph neural networks in conceptual cost estimation.

Keywords: Deep learning, Graph neural network, Conceptual cost estimation