MACHINE LEARNING MODEL FOR ESTIMATING THE COST OVERRUN OF PROCUREMENT MATERIALS IN CONSTRUCTION PROJECTS

Abstract

Procurement is a critical phase in construction projects, where inaccurate cost estimation can lead to significant cost overruns and reduced project efficiency. This study proposes a machine learning model based on an Artificial Neural Network (ANN) to predict procurement material cost overruns and identify the factors influencing procurement performance. The model was developed and evaluated using real-world procurement data collected from construction projects. Several factors related to estimation practices, stakeholder financial conditions, budgeting, government regulations, experience, and change orders were analyzed as model inputs. The results indicate that change orders and stakeholders’ financial conditions are among the most influential factors affecting procurement cost overruns. The proposed ANN model achieved a prediction performance exceeding 82%, demonstrating its capability to capture the complex relationships between procurement variables and cost outcomes. The developed model provides a practical decision-support tool for improving procurement planning, enhancing cost control, reducing financial risks, and supporting proactive management of potential changes during project execution.