Abstract

PurposeCurrent theoretical viewpoints regarding the performance trends of megaprojects endorse the notion that incorporating an outside perspective during the forecasting phase can be advantageous for the overall progress of the megaproject. This paper aims to propose a novel approach, the clustering-behavior analysis (C-BA), that leverages unsupervised machine learning to integrate an outside perspective as support for megaproject forecasting.Design/methodology/approachEmploying a database of 90 megaprojects, we demonstrated the application of C-BA. By utilizing unsupervised machine learning techniques, the method uncovers unforeseen patterns among past megaprojects, clusters them based on these patterns and allows for conducting a performance comparison with current megaprojects.FindingsThe findings reveal that the proposed C-BA method offers an effective alternative for supporting megaproject forecasting, aligning with the Fifth Hand principle. For practitioners, this would facilitate efficient benchmarking and has the potential to serve as a learning system within megaproject organizations.Originality/valueThe originality of this work lies in introducing a novel method that integrates an outside perspective into megaproject, with forecasts based on unsupervised machine learning. This approach aligns with the Fifth Hand principle and highlights the potential of artificial intelligence to serve as a learning system, offering a new avenue for efficient benchmarking in megaproject management. The paper adds complex network theory by giving the possibility of analyzing the uniqueness and unpredictable nature of megaprojects.