Temporal constraints are among the most significant challenges in construction project management, often causing schedule delays, cost escalations, quality decline, and dissatisfaction among stakeholders. This study presents an innovative integrated framework combining Structural Equation Modeling (SEM) and Machine Learning (Random Forest) to identify and predict the key factors affecting timely performance in construction projects. SEM results reveal that six critical factors—environmental conditions, stakeholders, regulations, policies, management, and traditional beliefs—have significant negative impacts on project performance. Among these, policies and management exhibit the strongest effects. The SEM model explains 67% of the variance in project outcomes, highlighting its explanatory power.

Following this, the Random Forest algorithm was employed to predict project timeliness, achieving an accuracy exceeding 87%, thus validating the importance of the identified factors. Feature importance analysis from the Random Forest model corroborates the findings from SEM, emphasizing the dominant role of policies, management, and regulations. This combined approach offers both a causal understanding and a powerful predictive capability, enabling early-stage identification of potential delays and risk factors.

The methodological novelty of this research lies in integrating SEM’s causal modeling with the predictive strength of machine learning, providing a comprehensive tool for construction project analysis. The findings hold practical implications for project managers and policymakers, especially in contexts like Iran where managerial and institutional challenges prevail. This framework supports data-driven decision-making and risk mitigation, ultimately aiming to improve the efficiency and success rates of construction projects.