Task-Based Risk Scoring for Early Prediction of Cost and Time Overruns in Construction Projects

Abstract

Construction projects often experience cost overruns and delay due to the cumulative effect of many risks happening in different activities. Regression-based forecasting, Monte Carlo simulation, and qualitative risk assessment are all well-established techniques, but their integration into a useful and transferable early-stage risk forecasting framework is still lacking. In order to convert expert risk assessments at the activity level into empirically calibrated project-level cost and duration multipliers, this study suggests a task-based risk scoring model. The study combines Monte Carlo simulation of project schedules and costs with expert-based qualitative risk identification organized using a standardized work breakdown structure. Power regression was used to create predictive relationships between baseline estimates and risk-adjusted outcomes using data from four multi-story building projects. Leave-One-Project-Out Cross-Validation (LOPOCV) was used to evaluate the robustness of the model, and Mean Absolute Percentage Error (MAPE) was determined to evaluate the accuracy, confirming low prediction error and strong explanatory capability. Dimensionless cost and duration scores are produced by the framework and can be immediately applied to baseline estimates. The findings show that while high-risk scenarios may increase 25% of project duration and 20% project cost, respectively, even low-risk scenarios may increase project duration and cost by roughly 6% and 7%.