A machine learning study to improve the reliability of project cost estimates
This study uses real data and earned value management metrics to develop an effective machine learning model for forecasting the cost throughout the project life cycle.
![](https://images.rennes-sb.com/frontend/2024/06/shutterstock_2261069627-scaled-e1718194190576.jpg)
Introduction
Today, in the new era of rapid technological developments, managers need to make quick decisions that deal with the uncertainties and complexities of dynamic business environments. Computational intelligence techniques of Artificial intelligence (AI) and data analytics can improve managerial decision-making under uncertainty. We address project management decision making in this study.
Purpose of the study
Project managers need reliable predictive analytics tools to make effective project intervention decisions throughout the project life cycle. Among several application areas of AI in project management, we address project cost forecasting and aim to predict the total project cost accurately. We use Machine learning (ML) to enhance reliability in project cost forecasting. We developed An XGBoost forecasting model and conducted computational experiments using real data of 110 projects representing 1268 cost data points. The projects were undertaken in Belgium and from the construction, information technology, and production industries.
The XGBoost model produces more accurate estimates at the early, middle, and late stages of the project execution, allowing for early warning signals for more effective cost control.
Results
We showed that the XGBoost model has a powerful learning effect and produces more accurate cost estimates. We improved the cost forecasting and specifically validated our methods on real data, which is rare in the literature. The model produces more accurate estimates at the early, middle, and late stages of the project execution, allowing for early warning signals for more effective cost control. In addition, it shows more accurate estimates in most projects tested, suggesting consistency when repeatedly used in practice.
Methodology
We developed an XGBoost algorithm-based model, which is a machine learning tool, to forecast the total cost of projects. We used the real cost data of 110 projects to validate the model. We compared the forecasting performance of the XGBoost model.
Applications and beneficiaries
Project managers can use our algorithm as a reliable predictive analytics tool to make effective project intervention decisions throughout the project life cycle.
Reference to the research
Timur Narbaev, Öncü Hazir, Balzhan Khamitova & Sayazhan Talgat (2024) A machine learning study to improve the reliability of project cost estimates, International Journal of Production Research, Volume 62 issue 12, pp. 4372-4388.