Civil Engineering Schedule Forensics: A Hybrid Delay Analysis Framework Using Machine Learning for Construction Projects

Abstract

This study created a hybrid framework that merged standard delay analysis methods with machine learning algorithms to make data-driven schedule forensics better. The study's goal was to increase the accuracy of identifying delays and estimating how long they will last in project plans by combining the domain expertise of traditional methods like Critical Path Method and Time Impact Analysis with the predictive power of machine learning models. The framework was trained and tested using historical data from a number of construction projects. The results showed that the hybrid strategy worked better than either classical or machine learning methods on their own for both classification and regression tasks. This made delay analysis more reliable and easier to understand. Feature importance analysis highlighted key factors influencing delays, while case studies confirmed the framework’s practical applicability. This study shows how hybrid data-driven methodologies could help improve forensic schedule analysis and risk management.