An Adaptive Neurofuzzy Inference System for the Assessment of Change Order Management Performance in Construction

Abstract

Change order management is a major challenge in the construction business due to the associated disputes, claims, productivity losses, delays, and cost implications. As a result, effective change order management (COM) is required to ensure the success of construction projects. The cost overruns and schedule delays caused by change orders have been recognized and researched by scholars and construction practitioners for decades. However, in modern construction management, there are additional performance factors that affect the performance of COM throughout construction activities. This study contributes to existing knowledge by identifying a comprehensive and multidimensional set of performance factors affecting COM and developing an adaptive neurofuzzy inference system (ANFIS) to model these factors quantitatively and evaluate COM implementation performance in the construction industry. Through an exhaustive literature search and engagement with specialists, 49 COM performance parameters were identified and then classified into seven groups. Then, 334 responses from building specialists were gathered via an online survey to determine the relative importance of each component and group. The obtained data were examined for normality, reliability, and independence and then analyzed using the Relative Importance Index (RII). The ANFIS model was constructed using a fuzzy clustering approach that took into account the clustering of input and output data sets, the fuzziness level of clusters, and the optimization of five Gaussian membership functions. The ANFIS model was subsequently validated using qualitative structural and behavioral testing (k-fold cross-validation). The findings of this study can be used as guidance in construction management for managing and evaluating the overall COM performance index of construction projects.