Predicting Schedule Delays and Cost Overruns in Iraqi Road and Bridge Projects Using Statistical Analysis and Explainable Artificial Intelligence
TL;DR
An integrated framework that combines statistical analysis, machine learning, and explainable artificial intelligence to identify, rank, and model project-risk factors in Iraqi transport infrastructure supports transparent early-warning assessment of financial, procurement, design, site, and management risks.
Abstract
Schedule delays and cost overruns remain persistent challenges in Iraqi road and bridge projects. This study presents an integrated framework that combines statistical analysis, machine learning, and explainable artificial intelligence to identify, rank, and model project-risk factors. The study uses a structured questionnaire with clients, consultants, contractors, project managers, planning engineers, cost-control personnel, and site engineers. At the same time, project-level records provide schedule, cost, variation, payment, design, procurement, and site variables. The analytical workflow includes descriptive statistics, the Relative Importance Index, Cronbach's alpha, exploratory factor analysis, correlation, group-comparison tests, regression, and machine-learning models including Ridge, LASSO, Decision Tree, Random Forest, XGBoost, Support Vector Regression, and a compact neural network. We evaluate predictive performance using MAE, RMSE, and R², with classification metrics where relevant, and use SHAP to support model interpretation. The numerical project and model results currently embedded in the manuscript are illustrative simulated outputs retained to demonstrate the proposed workflow. They must be replaced with verified project-level data before reporting empirical conclusions. The framework supports transparent early-warning assessment of financial, procurement, design, site, and management risks in Iraqi transport infrastructure