Evaluating the impact of uncertainty events on the cost of linear infrastructure projects

Abstract

This research examined the uncertainty events encountered in the process of constructing highways and evaluates their impact on South African highway construction costs. Due to the lack of appropriate evaluation of the impact of uncertainty events encountered in the construction process of highway projects, the construction costs of these projects are underestimated. To counteract such underestimation, this research developed an adaptive neuro-fuzzy inference system (Anfis) to assess the impact of uncertainty events involved in the construction of linear infrastructure projects. To validate the Anfis model, the stepwise regression analysis (SRA) and fuzzy Bayesian network (FBN) models were designed, and their results were compared with the outputs of Anfis. The prediction performance comparison proved that Anfis has a higher performance than SRA and FBN. Based on the results of the study, it can be deduced that the Anfis model is an accurate and reliable technique in assessing the impact of uncertainty events on the cost of construction projects. Therefore, the study concluded that using hybrid intelligent machine learning techniques not only minimises the time and difficulty of the estimation process but also reduces the potential inconsistency of correlation between variables in construction cost prediction.