Assessing cyber risks in construction projects: A machine learning-centric approach

Abstract

The construction industry is undergoing digitalization, but it is increasingly vulnerable to cyber attacks due to its slow pace in developing effective cyber risk assessment tools. This study develops a Machine Learning (ML)-centric approach to assess common cyber risks for construction projects. This approach comprises three components: (1) For risk prediction, a simulated dataset is generated using Monte Carlo simulations, which is utilized for model training. A two-phase model development strategy is proposed to select the optimal model for each risk. (2) For risk factor analysis, ML feature analysis methods are adapted to identify risk factors that contribute significantly to risks of specific projects. (3) For the risk reduction strategy, a greedy optimization algorithm is proposed to efficiently address high-contributing risk factors. To demonstrate the applicability of the developed approach, a case study is conducted on a real construction project.