In the ever-evolving landscape of construction project management, predicting cost contingencies has long been a challenge fraught with uncertainty. However, a groundbreaking study led by Acinia Nindartin from Samsung C&T’s Engineering & Construction Group in Seoul, Republic of Korea, is set to revolutionize this aspect of the industry. Published in the *Journal of Civil Engineering and Management* (translated from Lithuanian as *Civilinė Inžinerija ir Vadyba*), the research introduces machine learning (ML) algorithms to enhance the accuracy of cost contingency predictions, offering significant commercial impacts for the energy sector and beyond.
Construction projects are inherently dynamic, with costs often deviating from initial estimates due to unforeseen changes and uncertainties. Traditional methods of predicting cost contingencies have often fallen short in reliability and accuracy. Nindartin’s research aims to address this gap by leveraging the power of machine learning. “The goal was to develop a model that could more accurately predict cost contingencies, thereby helping project managers allocate budgets more effectively and reduce financial risks,” Nindartin explained.
The study utilized datasets from construction transportation projects bid between 2013 and 2017, collected from the Florida Department of Transportation (FDOT) website. To tackle the issue of imbalanced regression datasets, the research introduced the synthetic minority over-sampling technique for regression with Gaussian noise (SMOGN) algorithm. This innovative approach helped balance the data, improving the performance of the machine learning models.
Among the various ML algorithms tested, the random forest (RF) regression model, coupled with random search hyperparameter optimization, emerged as the most accurate predictor. It outperformed other models like extreme gradient boosting (XGBoost) regression and artificial neural network (ANN) models. The research also identified four key parameters that significantly influence cost contingency predictions: project amount, project duration, and the geographical factors of latitude and longitude.
The implications of this research are profound for the construction and energy sectors. Accurate prediction of cost contingencies can lead to more precise budgeting, reduced financial risks, and improved project management. “This model provides a more advanced method for predicting cost contingencies, which can be a game-changer for project managers and stakeholders,” Nindartin noted.
As the construction industry continues to embrace digital transformation, the integration of machine learning algorithms like those explored in Nindartin’s study could become a standard practice. This shift not only enhances the efficiency and accuracy of cost predictions but also paves the way for more innovative and data-driven decision-making processes.
The findings of this research, published in the *Journal of Civil Engineering and Management*, offer new insights for both researchers and practitioners. By adopting these advanced methods, the construction industry can move towards more predictable and financially secure project outcomes, ultimately benefiting the broader energy sector and other related industries.