A Predictive Analytics Model for Early Detection of Budget Overruns in Large-Scale Projects Using Integrated Financial and Operational Data

Abstract

Budget overruns remain a persistent challenge in large-scale projects, particularly in sectors such as construction, oil and gas, and information technology, where complexity and uncertainty significantly influence cost performance. Traditional cost control models, including Earned Value Management (EVM), rely primarily on retrospective indicators, limiting their ability to provide early warning signals for emerging financial risks. This study proposes a predictive analytics framework for the early detection of budget overruns by integrating financial and operational data within a mathematically grounded modelling structure.

The methodology combines feature engineering, regression-based forecasting, and time-series modeling to estimate future cost trajectories. Key variables, including actual cost, schedule deviation, and resource utilization, are transformed into predictive features to capture dynamic project behavior. A novel Budget Overrun Risk Index (BORI) is introduced to quantify the extent of predicted cost deviation relative to the approved budget, enabling standardized risk assessment across projects. The model is evaluated using statistical performance metrics such as Root Mean Square Error (RMSE) and the coefficient of determination (), alongside cross-validation techniques to ensure robustness and generalization.

Results demonstrate that the proposed model significantly improves prediction accuracy compared to traditional EVM approaches, with lower error rates and higher explanatory power. Importantly, the framework enables the identification of cost deviation signals before 30–40% project completion, thereby extending the forecasting horizon and supporting proactive decision-making. Sensitivity analysis further confirms the impact of cost growth rate and schedule delays as dominant drivers of budget overrun risk.

The study concludes that integrating financial and operational datasets enhances model performance and provides a more comprehensive understanding of project dynamics. While challenges related to data quality and model complexity persist, the proposed framework offers a scalable and interpretable solution for real-time project monitoring. The findings contribute to the advancement of predictive analytics in project management and provide a foundation for future research on hybrid AI-driven cost control systems.