PREDICTIVE MODELING OF CONSTRUCTION PROJECT HEALTH USING MACHINE LEARNING–BASED MULTIVARIATE INDICATORS

Abstract

Construction projects are inherently complex and prone to delays, cost overruns, safety issues, and quality deviations, making timely monitoring of project health critical for successful delivery. This study investigates the use of machine learning–based predictive modeling to assess and forecast construction project health using multivariate performance indicators. A hypothetical dataset comprising multiple project dimensions—including cost, schedule, quality, safety, resource utilization, and stakeholder coordination—was analyzed to construct a Project Health Index, categorizing projects as healthy, moderately at-risk, or critical. Various machine learning algorithms, including logistic regression, decision trees, random forest, support vector machines, gradient boosting, and artificial neural networks, were employed to predict project health. The results indicate that ensemble and nonlinear models, particularly gradient boosting, provided the highest prediction accuracy, demonstrating the complex interdependencies among performance indicators. Findings highlight the effectiveness of data-driven predictive approaches in enabling early detection of project risks, supporting proactive decision-making, and enhancing overall project performance. The study underscores the potential of integrating AI-driven predictive systems into construction project management practices for improved governance and operational resilience.