Artificial Intelligence in Construction Management Enhancing Decision Making A cross a Project Life Cycle: throught A Machine Learning Method

Abstract

The construction industry, which creates 13% of the world’s GDP, suffers from cost overruns of 28% and delays of over 42%, yet is still undigitized. AI has the promise of transforming management in the areas of cost, time and safety. This research provides a systematic review of AI applications and performs an original machine learning analysis on seven algorithms, i.e., Support Vector Machines, Logistic Regression, K-Nearest Neighbors, Random Forest, Stacking Ensemble, Artificial Neural Networks, and XGBoost on 139 Indian road projects with 44 risk factors. XGBoost performed the best with 82.50% accuracy and 84.21% F1-score. The most important concerns were the delay in decision-making (0.152), cash flow problems (0.082) and cost underestimate (0.071). Case studies demonstrate AI’s capabilities: 98.7% accurate cost estimation, 94.36% delay prediction via deep learning, and 97% safety monitoring with computer vision. These findings underscore the importance of ensemble ML in proactive, data-driven risk management, bringing projects into a predictive framework rather than reactive.