Journal of Resource Management and Decision Engineering In Press 2027 11 01 Predicting Costs and Risks in Construction Projects Using BIM and Artificial Intelligence Data Analytics Predicting Costs and Risks in Construction Projects Using BIM and Artificial Intelligence Data Analytics 1 15 EN Journal Article 2026 02 01 <p class="Abstract" style="line-height: normal;"><span lang="EN-GB" style="letter-spacing: -.2pt;">This study aimed to predict cost overrun and project risk in construction projects using BIM-based indicators and artificial intelligence data analytics. This applied, quantitative, predictive, and analytical study was conducted on 64 construction projects in Tehran, Iran, with the participation of 148 construction professionals, including project managers, BIM specialists, cost and quantity surveying experts, planning and project control engineers, and site engineers. Data were collected using BIM-based project documentation, cost and schedule records, risk assessment forms, and expert validation checklists. The main variables included BIM maturity level, number of detected BIM clashes, design revision frequency, schedule deviation, material cost fluctuation, procurement delay, change orders, rework cost ratio, cost overrun percentage, and composite project risk score. Data were analyzed using correlation analysis and artificial intelligence-based predictive models, including multiple linear regression, decision tree, support vector regression, artificial neural network, random forest, and gradient boosting models. Inferential findings showed that cost overrun had significant positive correlations with schedule deviation, design revisions, BIM clashes, material cost fluctuation, and composite project risk score, while BIM maturity had significant negative correlations with cost overrun, schedule deviation, BIM clashes, and risk score. Gradient boosting regression produced the strongest cost prediction performance, with the lowest mean absolute error and root mean square error and the highest coefficient of determination, explaining 84% of the variance in cost overrun. For project risk classification, gradient boosting also achieved the highest performance, with accuracy of 0.89 and area under the curve of 0.93. Feature importance analysis identified schedule deviation, design revision frequency, BIM clashes, material cost fluctuation, and BIM maturity as the strongest predictors of cost and risk outcomes. The findings indicate that integrating BIM-based project information with artificial intelligence analytics provides a reliable predictive framework for early identification of cost overrun and project risk in construction projects. AI-based models, especially gradient boosting, can support proactive decision-making, improve cost control, and strengthen risk management throughout the project life cycle.</span></p> https://www.journalrmde.com/index.php/jrmde/article/download/381/352