Evaluating the impact of construction delays on project duration using machine learning and multi-criteria decision analysis

Abstract

Construction projects are inherently complex and prone to delays, significantly impacting project timelines and costs. This study addresses the critical issue of construction delays in Jordan by leveraging advanced methodologies such as Gaussian Process Regression (GPR) and the Analytical Hierarchy Process (AHP). The problem of accurately predicting and managing delays in construction projects has long challenged the industry, with existing approaches often failing to account for the multifaceted nature of delay factors. This research integrates GPR, a machine learning technique, with AHP, a Multi-Criteria Decision Analysis (MCDA) tool, to evaluate and predict the impact of delay factors on project duration. The study employs a comprehensive dataset comprising 191 construction projects in Jordan, with critical variables identified through expert evaluations and literature reviews. The GPR model demonstrated superior predictive capabilities, achieving an R² value close to 1, indicating its high accuracy in forecasting time and cost overruns. The AHP model, on the other hand, prioritized weather conditions and unrealistic contract requirements as the most significant contributors to delays. The findings suggest that the combined application of GPR and AHP offers a robust framework for predicting and managing construction delays, providing valuable insights for improving project management practices. Future work should focus on expanding the dataset and refining the models to enhance their applicability across different regions and project types.